ESSAYS ON MANAGING MANUFACTURING AND SERVICE NETWORKS

By

Seongwon Park

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

Business Administration - Operations and Sourcing Management—Doctor of Philosophy

2025

ABSTRACT

This dissertation investigates how network structures and relational dynamics shape environ-

mental and operational outcomes, with a focus on supply chain systems in manufacturing and

service industries. The first two essays examine manufacturing supply networks, emphasizing the

role of interfirm relationships in influencing environmental performance and transparency. The

first essay explores the impact of downstream firms on the environmental efficiency of upstream

suppliers across multi-tier supply chains. Drawing on FactSet Revere and Trucost data from six

manufacturing sectors, the analysis reveals that downstream firms’ environmental efficiency pos-

itively influences that of upstream partners – a relationship that intensifies under higher levels of

financial dependence.

The second essay examines the effect of buyer environmental disclosure on supplier disclosure,

addressing concerns about potential supply chain leakage. Using data from FactSet Revere and

Bloomberg environmental, social, and governance (ESG) covering the materials and pharmaceu-

ticals sectors, the study finds no consistent evidence of a broad negative effect, suggesting that

leakage may not be prevalent under voluntary disclosure regimes. However, the findings high-

light key moderating dynamics. Structural social capital, reflected in shared network ties, plays

a significant role: buyer–supplier pairs with weaker structural overlap are more likely to exhibit

negative effects, while stronger overlap mitigates this risk. Similarly, cultural similarity moderates

outcomes – greater cultural dissimilarity correlates with negative effects on supplier disclosure,

whereas culturally aligned pairs show no significant effect.

The third essay turns to service networks, specifically examining how airline network configu-

rations influence the management of operational disruptions. Using data from the U.S. Bureau of

Transportation Statistics across seven major U.S. airlines, the analysis demonstrates that the network

structure of an airline plays a critical role in managing delays, cancellations, and baggage handling

issues. Together, these three essays contribute to a deeper theoretical and practical understanding

of how varying network contexts – across both manufacturing and service domains – affect firms’

environmental and operational performance.

ACKNOWLEDGEMENTS

I am deeply grateful to all those who have supported and encouraged me throughout the course

of my doctoral studies. At the forefront of this journey stands my advisor, Dr. Srinivas (Sri) Talluri,

to whom I owe my deepest appreciation. His support has been unwavering and his guidance

consistently thoughtful. His mentorship not only enabled me to reach this important milestone

but also shaped the intellectual foundations of my work. I am especially appreciative of his deep

expertise, his generosity with time, and his steadfast commitment to my development. Working

under his guidance has been both a privilege and an inspiration.

I also wish to express my sincere appreciation to my committee members: Dr. Sriram

Narayanan, Dr. Adrian Choo, and Dr. Chandrasekhar Manchiraju.

I am especially grateful to

Dr. Narayanan for his insightful teaching in his seminar and for his thoughtful guidance, both of

which laid a strong academic foundation for my scholarly development. I am equally indebted to

Dr. Choo, whose dual role as a doctoral program director and colleague helped me grow not only

as a researcher but also as a more thoughtful and considerate individual. I would also like to thank

Dr. Manchiraju for his thoughtful guidance and encouragement throughout the development of

this dissertation. His academic rigor and dedication have been a valuable source of motivation in

shaping my own approach to research.

Lastly, I am deeply grateful to my family. My mother and brother have offered steadfast

support for my academic pursuits, and their encouragement has been a constant source of strength

throughout this journey. I also wish to express my heartfelt gratitude to my father in heaven, whose

love and guidance inspired me throughout this journey. Most of all, I am deeply thankful to my

wife, Jiin Huh, whose constant love, patience, and encouragement have supported me at every stage

of this journey. Her steady belief in me has been both my greatest motivation and the foundation

of my perseverance.

iii

TABLE OF CONTENTS

CHAPTER 1

INTRODUCTION .
.

Introduction . . . . . . .

.
.

.
.

.
.

.
.

.
.

.
.

.
.
. .

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

1.1

CHAPTER 2

.

.

DYNAMICS OF ENVIRONMENTAL MANAGEMENT
.
IN MULTI-TIER SUPPLY CHAINS .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
Introduction . . . . . . .
2.1
2.2 Background and Hypotheses .
.
2.3 Research Design . . . . .
.
.
2.4 Analysis and Results
. .
.
.
.
2.5 Sensitivity Analysis . . .
.
2.6 Discussion and Conclusion .
.
.
REFERENCES . . . . . . . .
.
.
APPENDIX . . . . . . . . . .

. .
.
. .
. .
. .
. .
. .
. .

.
. .
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.

.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

.
.
.
.
.
.
.
.
.

.
.

1
1

4
.
4
.
.
8
. 12
. 25
. 30
. 31
. 36
. 44

.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.

. .
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.

.
. 47
.
. 47
. 52
.
. . 56
. 67
.
. 79
.
. 85
.
. 94
.
. 103
.

.
.
.
.
.
.
.
.
.

. 116
. 116
. 123
. 129
. 134
. 140
. 146
. 150
. 155

CHAPTER 3

.
.

.
.

.
.

.
.

.
.

.
.

3.1
Introduction . . . . . . .
3.2 Literature Review . . . .
3.3 Hypotheses Development and Conceptual Framework .
.
.
3.4 Data Source . . . . . . .
.
.
.
3.5 Empirical Specification . .
.
3.6 Robustness Checks and Additional Analysis .
.
.
3.7 Discussion and Conclusion .
.
.
.
REFERENCES .

ENVIRONMENTAL DISCLOSURE IN SUPPLY NETWORKS .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
. . .
. .
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.

. .
.
.
.
.
.
.
.
.

. . . .

.
.
.
.
.
.
.
.

.
.
. .

.
.
. .

.
.
. .

. .
. .

. .
. .

. .
. .

. . .

.
.
.
.
.

.
.
.
.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.
.

.

CHAPTER 4

EFFECTIVENESS OF ALTERNATIVE NETWORK
CONFIGURATIONS IN THE AIRLINE INDUSTRY .
. .
.
.
.
.
.
.
. . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.

.
.
Introduction . . . . . . .
4.1
.
4.2 Literature Review . . . .
.
.
4.3 Hypotheses Development .
4.4 Data . . . . . . . . . . .
.
.
4.5 Empirical Specification & Results
.
4.6 Additional Analysis . .
.
.
4.7 Discussion . . . . . . . .
.
. . . .
REFERENCES .

.
.
. .
.
.
.
.
.
.
. .
.
.
.
.

.
.
. .
.
.
.
.
.
.
. .
.
.
.
.

.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
. .
.
.
.
.

.
.
.
.
. .
.
.
.

. .
. .
. .
. .
.
. .
. .
. .

.
.
.
.
.
.
.
.

. . .

.
.
.
.

.
.
.
.

.
.
.

.
.
.

.
.
.

.
.
.

.
. .
.
.
.
. .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

iv

CHAPTER 1

INTRODUCTION

1.1

Introduction

With the global expansion of business across manufacturing and service sectors, firms are

increasingly expected not only to manage their own performance but also to engage with a diverse

set of partners throughout the supply chain. As inter-firm relationships become more complex

and interconnected, the scope of organizational responsibility extends beyond direct interactions to

encompass indirect network influences. Paradoxically, increasing interdependence within global

business networks often places greater operational and relational demands on the firms involved.

This dissertation examines how both inter-organizational and intra-organizational networks affect

firm performance, with a specific focus on corporate sustainability practices and operational disrup-

tions in the service industry. By integrating insights from supply chain management, sustainability,

and service operations, this work contributes to the growing literature on how network structure

and relational dynamics shape organizational outcomes in both manufacturing and service sectors.

Spanning three empirical essays, this dissertation addresses underexplored but critical questions

surrounding environmental management, corporate sustainability reporting, and operational relia-

bility in networked contexts. The first essay investigates how downstream firms in multi-tier supply

chains (MSCs) affect the environmental efficiency of their upstream partners. Using FactSet Revere

supply chain relationship data and Trucost environmental performance data across six manufactur-

ing industries, the study constructs firm-level environmental efficiency scores with network data

envelopment analysis (NDEA). It further examines how suppliers’ revenue dependence on down-

stream buyers moderates this cross-tier environmental effect. Drawing on 259 triadic relationships

from 2014 to 2018, the analysis applies nonlinear models to explore the environmental influence

across supply chain tiers. Given the lack of prior studies on cross-tier sustainability activities using

secondary data, this study will provide useful insights into the dynamic interplay between firms in

managing cross-tier environmental efficiency based on the use of a mixed-methods approach and

the unique panel data set.

1

The second essay explores environmental disclosure behavior within buyer–supplier relation-

ships (BSRs), particularly in the context of the supply chain leakage phenomenon and the pollution

haven hypothesis. Using dyadic relationship data from FactSet Revere and Bloomberg environ-

mental, social, and governance (ESG) data covering two manufacturing sectors over a five-year

panel (2015–2019), the study finds no broad evidence of a negative influence of buyer disclosure

on supplier environmental disclosure. However, the findings reveal important moderating effects.

Structural social capital, represented by network overlap between buyers and their suppliers, and

cultural similarity influence the extent to which buyers shape their suppliers’ environmental disclo-

sure behavior. To be specific, weak structural overlap or cultural dissimilarity amplifies the negative

effect, while strong overlap or similarity mitigates it. Relationship duration, by contrast, does not

significantly moderate the effect. Taken together, these findings enhance theoretical and managerial

understanding by illustrating how different dimensions of social capital can differentially influence

environmental practices within supply networks.

The third essay shifts the focus from supply chains to service networks, examining how dif-

ferent airline network configurations, hub-and-spoke versus point-to-point, affect service quality

in the U.S. domestic market. Although network structure is widely recognized as a key determi-

nant of airline performance, prior research has not systematically investigated its role in shaping

operational resilience during disruptions. Using flight-level data from the U.S. Bureau of Trans-

portation Statistics (BTS), this study compares the performance of full-service carriers (FSCs) and

low-cost carriers (LCCs) through both linear and nonlinear modeling approaches. The results indi-

cate that FSCs, which primarily operate hub-and-spoke networks, outperform LCCs in managing

delays, reducing cancellations, and maintaining higher on-time performance. By contrast, LCCs

demonstrate stronger performance in baggage handling. The analysis also reveals notable variation

within the LCC group. In particular, Southwest Airlines, despite following a point-to-point model,

performs well in reducing delays and baggage incidents but exhibits a higher rate of flight cancel-

lations. These findings offer important insights into how network configurations shape operational

outcomes in service industries.

2

Collectively, the three essays provide both theoretical and empirical insights into how network

structures and relational dynamics shape firm behavior and performance. Rather than focusing

solely on dyadic relationships, this dissertation examines the broader systemic and cross-tier inter-

actions that define modern supply chains. It further extends this perspective to service operations

by analyzing how network configurations influence operational performance and service reliability

in the airline industry. By leveraging large-scale secondary data and applying a combination of

linear and nonlinear modeling techniques, the dissertation highlights the importance of designing

and managing networks not only to improve efficiency but also to enhance resilience and support

sustainability.

3

CHAPTER 2

DYNAMICS OF ENVIRONMENTAL MANAGEMENT
IN MULTI-TIER SUPPLY CHAINS

2.1

Introduction

In recent years, several countries have entered into international environmental treaties, such as the

Kyoto Protocol, the Montreal Protocol, and the Paris Agreement, to help reduce the adverse impact

caused by humans on the environment. Additionally, government bodies, such as the Environmental

Protection Agency (EPA) of the United States, establish sustainability initiatives that assist firms

in enhancing the sustainability of their supply chains through assessment, benchmarking, and

enhancement of operational efficacy. SmartWay is an example of a voluntary public-private

program introduced by the EPA in 2004 that has helped firms advance supply chain sustainability

by reducing freight emissions. In addition to governments, non-profit organizations such as the

Carbon Disclosure Project (CDP) help firms disclose their environmental impact and develop

carbon emissions reduction strategies. According to a 2019 report (CDP, 2019) by the CDP, the

collaborative action between the CDP Supply Chain program members and their suppliers reduced

8 million tonnes of carbon emissions.

With the rising demand for transparency in supply chains, focal companies and their suppliers are

being held responsible for sustainability challenges within their supply chains (Wilhelm et al., 2016a;

Awaysheh and Klassen, 2010). Accordingly, businesses are striving to reduce their environmental

and social impacts (Bové and Swartz, 2016). Many multinational corporations have set social

and environmental standards to manage suppliers in addition to pursuing social and environmental

goals of their own (Villena and Gioia, 2020). However, from the viewpoint of focal companies in

multi-tier supply chains (MSCs), managing the risk of environmental and social breaches caused

by lower-tier suppliers located beyond direct suppliers is more challenging (Wilhelm and Villena,

2021; Villena and Gioia, 2018). This is because focal firms’ control over lower-tier suppliers tends

to be lower given the lack of information on their sustainability activities, the global nature of the

supply chain (Wilhelm et al., 2016b), and distant relationships (Bellamy et al., 2020; Awaysheh

4

and Klassen, 2010). According to a 2022 survey by Ernst and Young (Ernst & Young, 2022),

over 60% of supply chain managers report no year-over-year enhancements in the comprehensive

visibility of their supply chains. Thus, lower-tier suppliers are more passive in addressing social

and environmental sustainability challenges as they perceive a low risk of being penalized for not

attending to these challenges (Villena and Gioia, 2018). Although numerous global conglomerates

have been incorporating sustainability practices into their business operations, the impact of these

practices on the sustainable operations of upstream partners remains unclear and warrants further

investigation (Hartmann and Moeller, 2014).

In this paper, our primary research question is to examine the cross-tier environmental impact

on firms in MSCs resulting from the environmentally responsible actions of their downstream

partners. Answering this question is challenging due to the lack of available data on upstream

suppliers and the shortage of related studies in the literature (Villena and Gioia, 2018). To address

this, we primarily use FactSet Revere supply chain relationships data in this study. Under an

open triadic configuration, we consider numerous triadic relationships from the viewpoint of focal

companies engaged in the general category of manufacturing. We augment the FactSet data

using Trucost environmental data, which provides information on hundreds of key environmental

performance indicators across 500 distinct industry sectors. Since MSCs involve multiple firms at

each tier, we aggregate the environmental performance data and create comprehensive measures for

each triadic configuration using network data envelopment analysis (NDEA). The environmental

efficiency scores derived from NDEA provide a holistic approach to assessing firms’ environmental

performance across supply chains, as it uses multiple input and output factors based on both

desirable and undesirable outcomes. Moreover, using NDEA in studies involving diverse supply

chain activities is practical because it captures both the internal and external activities of firms that

comprise the supply chains. This procedure helps us address whether the environmental efficiency

of downstream firms positively influences the efficiency of not only their neighboring suppliers but

also distant suppliers.

Our second research question is to enhance the understanding of how the cross-tier environmen-

5

tal impact in buyer-supplier relationships (BSRs) is strengthened by supplier revenue dependence.

To achieve this, we revisit power-dependence relations from a supplier’s perspective in the context

of sustainable supply chain management (SSCM). While organizational performance is contingent

on that of its partner, a firm often encounters a dilemma as its dependence on another organization

becomes more interlocked (Pfeffer and Nowak, 1976). Thus, the performance benefits of relational

dependence between firms may vary depending on the degree of the actual involvement in the

relationship. Furthermore, addressing environmental concerns within supply chains is a shared pri-

ority, which, in turn, enables supply chain partners to support each other’s operations and enhance

their environmental sustainability (Hajmohammad et al., 2024; Kim and Henderson, 2015; Kim and

Wemmerlöv, 2015; Hoejmose et al., 2013; Gimenez and Sierra, 2013). Consequently, in light of the

environmental impact stemming from neighboring downstream partners on an upstream partner’s

environmental efficiency, we also address the moderating effect of supplier revenue dependence on

the environmental association.

2.1.1 Contributions and Results

Our research contributes to the SSCM literature in three important ways as discussed below.

(i) Impact beyond first-tier: A majority of the research in SSCM concentrates on investigating

the impact of focal firms on first-tier suppliers, and vice versa (Diebel et al., 2024; Song

et al., 2024, 2023; Jira and Toffel, 2013). This perspective inherently assumes that the first-

tier suppliers will oversee the lower tiers of supply chains in a manner aligned with and

conducive to the objectives of the focal company. However, there is evidence indicating

the ineffectiveness of this approach, as it puts the focal company at risk of dealing with

and controlling adverse impacts resulting from the actions of the upstream members of their

supply chains (Choi and Linton, 2011). Thus, to investigate the environmental impact of

downstream firms on both the closest and distant suppliers, it is essential to adopt a holistic

approach that evaluates performance beyond a dyadic level. Hence, in our study, we consider

triadic relationships to precisely capture the internal dynamics in MSCs.

6

(ii) Multiple criteria and network impact in measuring environmental efficiency: Many

studies in the SSCM literature consider only a single criterion to capture environmental per-

formance (De Stefano and Montes-Sancho, 2024; Song et al., 2024; Adhikary et al., 2020).

However, industrial production and business operations require diverse input resources and

generate multiple negative environmental externalities. Thus, from a managerial viewpoint,

it is necessary to consider a comprehensive measure that accounts for multiple desirable and

undesirable environmental outcomes. We address this using an NDEA model, which creates

a comprehensive measure of environmental efficiency considering multiple inputs and out-

puts of firms’ operations that impact the environment. Moreover, our study contributes to

the literature by employing a network DEA model that effectively captures the interrelation-

ships between different entities within supply chains. This approach allows us to compute

environmental efficiency scores that reflect the network interrelationships of firms.

(iii) Time-varying impact and revenue dependence: Many studies in the literature examine

the environmental impact between members of MSCs over a limited time period. However,

it is important to note that inter-firm environmental impact may vary over time.

In our

study, we utilize a unique panel dataset that allows us to conduct a robust analysis by

capturing time-dependent measures of environmental efficiency. This enables us to test

the inter-firm environmental impact not only across multiple firms but also over different

periods. Moreover, our study contributes to the literature on power-dependence relations

within SSCM by examining the effect of the revenue dependence of upstream suppliers (on

their downstream firms) on the environmental impact between members of MSCs.

We employ a mixed methods approach, integrating the environmental efficiency scores derived

from our NDEA model as variables into our econometric models to test our research hypotheses.

Our main findings are as follows: (i) The environmental efficiency of focal firms positively influ-

ences the efficiency of their first-tier suppliers. (ii) The environmental efficiency of second-tier

suppliers is impacted by the first-tier suppliers but not directly by the focal firms, suggesting that

7

second-tier suppliers are more likely to align with their closest downstream partners. (iii) At any

level of environmental efficiency, the downstream firms have a similar influence on the efficiency of

their upstream partners. Moreover, the environmental influence of focal firms on first-tier suppliers

is greater than the influence of first-tier suppliers on second-tier suppliers. (iv) Supplier revenue

dependence enhances the positive relationship between the environmental efficiencies of MSC

members, suggesting that higher revenue dependence of upstream partners on downstream firms in

BSRs improves the impact of downstream firms on upstream partners’ environmental efficiency.

2.1.2 Organization

The rest of our paper is structured as follows.

In the next section, we review the related

literature on environmental management in the context of MSCs and develop our hypotheses.

Based on a thorough investigation of the contextual background, Section 3 presents our research

design, including details on supply network configurations, data and sources, our NDEA model

developed using diverse environmental performance measures, and the definitions of our variables.

We present our estimation models and a discussion of our results in the fourth section. In Section 5,

we offer additional insights through sensitivity analyses based on the results of nonlinear fractional

probit models. Finally, in Section 6, we discuss the implications of our findings and provide

directions for future research.

2.2 Background and Hypotheses

In this section, we discuss the relevant literature in SSCM and develop our hypotheses. We

first develop our hypotheses related to cross-tier environmental impact in MSCs, followed by our

hypotheses on the moderating effect of supplier revenue dependence on this impact.

2.2.1 Relational Dynamics in Multi-tier Supply Chains

Sustainability has been subsumed in the literature of diverse business disciplines, including

operations and supply chain management (SCM) (Kleindorfer et al., 2005). Environmental ac-

countability and transparency in supply chains are becoming increasingly important for firms (Song

et al., 2024, 2023; Gualandris et al., 2021). Researchers, however, indicate a lack of knowledge re-

garding the tracing and measuring of social and environmental impacts and their externalized costs

8

from global production and consumption systems (O’Rourke, 2014). This is because global supply

chains have become more complex than ever due to the expansion of supply chain boundaries, in

addition to firms being challenged with managing the depth and quality of BSRs and partners’

malpractices that might harm their operations and reputation. Further, despite the increase in

transparency and accountability in globalized supply chains, consumers do not differentiate the

efforts made by members in MSCs and attribute the unsustainable behavior of suppliers to focal

firms; this is referred to as “chain liability” (Hartmann and Moeller, 2014; Tachizawa and Wong,

2014). Therefore, understanding cross-tier environmental impact in MSCs is essential.

Given the nature of MSCs, member firms inevitably affect each other. Firms can enhance their

productivity when they collaborate with productive partners by adopting their partners’ efficient

practices. Thus, a strong and productive partner’s knowledge spills over through different channels

across the supply chains (Serpa and Krishnan, 2018). This indicates that the environmental

efficiency of productive partners could potentially transfer to the other members of the supply chain.

In addition, when focal firms establish direct and contractual relationships with first-tier suppliers,

they sometimes enforce compliance with sustainability measures in the contractual arrangements to

control their own sustainability performance (Saunders et al., 2020; Hartmann and Moeller, 2014).

To promote the environmental stewardship of supply networks, some focal firms also leverage a

mimetic mechanism (Diebel et al., 2024) in which the suppliers mimic the sustainable practices of

the focal firm. Consequently, it is highly likely that first-tier suppliers in MSCs are more receptive

to the sustainability practices of focal firms. Hence, we hypothesize the following:

Hypothesis 1A (H1A). The environmental efficiency of focal firms positively influences the

environmental efficiency of first-tier suppliers.

Previous studies have shown environmental influence not only between a focal firm and first-tier

suppliers but also between first-tier suppliers and lower-tier suppliers; see e.g., Jamalnia et al. (2023),

Wilhelm et al. (2016b), Tachizawa and Wong (2014), Touboulic et al. (2014). First-tier suppliers

often serve as a boundary-spanner to improve sustainability performance across supply chains

(De Stefano and Montes-Sancho, 2024; Grimm et al., 2014; Touboulic et al., 2014). Moreover,

9

first-tier suppliers often monitor the sustainability performance criteria of second-tier suppliers and

require them to meet these criteria to mitigate potential social and environmental risks (Wilhelm

and Villena, 2021; Wilhelm et al., 2016b). Thus, we posit the following:

Hypothesis 1B (H1B). The environmental efficiency of first-tier suppliers positively influences

the environmental efficiency of second-tier suppliers.

The impact of focal firms on the sustainable practices of suppliers beyond the first tier is not

straightforward. Lower-tier suppliers often proactively adopt environmentally conscious practices

when they are more exposed to the actions of their downstream customers who are sustainability

leaders conscious of environmental issues. They tend to actively participate in their customers’

industry organizations and imitate their sustainability initiatives (Villena and Gioia, 2018). Ad-

ditionally, the intermediaries who oversee lower-tier suppliers’ operations on behalf of the focal

firms adopt helpful procedures to aid lower-tier suppliers in fulfilling their sustainability obligations

(Jamalnia et al., 2023).

In contrast, studies in the literature also argue that the sustainable actions of focal firms may

not impact suppliers beyond the first tier. Since focal firms can directly monitor the first-tier

suppliers, the closest partners strive to meet sustainability standards, whereas lower-tier suppliers are

sometimes too passive to engage in environmental and social issues. Managing the sustainability of

distant partners is more complex and less controllable since they often have no direct and contractual

relationship with focal firms. Instead, they indirectly communicate with focal firms with the help of

intermediaries (Jamalnia et al., 2023; Hartmann and Moeller, 2014). Although focal firms delegate

their responsibility to intermediaries for overseeing lower-tier suppliers and managing sustainability

considerations of their upstream processes (Wilhelm et al., 2016a,b), lower-tier suppliers are still

less likely to engage in social and sustainability obligations imposed by focal firms (Jamalnia

et al., 2023; Villena and Gioia, 2018). Thus, in the absence of a contractually binding force, it is

difficult for focal firms to regulate lower-tier suppliers and oversee their violations of sustainability

requirements. This leads to lower-tier suppliers’ perception that they are less likely to be penalized

for not addressing environmental issues (Villena and Gioia, 2018), potentially leading to negative

10

environmental consequences and stakeholder reactions for the focal firms (Jamalnia et al., 2023).

Moreover, as the physical and organizational separation between focal firms and upstream partners

expands, the focal firms experience difficulties in closely interacting with lower-tier suppliers,

leading to challenges in data gathering, which results in a lack of visibility (Garcia-Torres et al.,

2019; Wilhelm et al., 2016a; Awaysheh and Klassen, 2010). Despite focal firms becoming equipped

with the ability to track and address social and environmental issues across supply chains due to

technological advancements (Garcia-Torres et al., 2019), their awareness of the identities of lower-

tier suppliers and their ability to directly influence them remain limited (Grimm et al., 2014). Thus,

we hypothesize the following:

Hypothesis 2 (H2). The environmental efficiency of focal firms directly influences the envi-

ronmental efficiency of second-tier suppliers.

2.2.2 Moderating Effects of Supplier Financial Dependence on Environmental Performance

Inter-organizational relationships have often been conceptualized via the resource dependence

theory (RDT), which takes into account inter-relational issues such as dependence, uncertainty,

power, and scarcity of resources (Hajmohammad et al., 2024; Tate et al., 2022). From a resource-

dependence perspective, despite reducing external reliance leading to greater success, firms are

typically not self-sufficient and rely on their partners to acquire necessary resources (Elking et al.,

2017; Pfeffer and Salancik, 1978). Thus, inter-organizational relationships can be configured dif-

ferently based on the diverse power dynamics of the partners involved. In SCM, supply networks

comprise numerous organizations that are intricately connected and engaged in operational activi-

ties. Consequently, in MSCs, interdependence can serve as a driving force for fostering cooperative

sustainable relationships among involved entities (Tate et al., 2022).

Dependence in relationships can also be found in the context of sustainability. The diffusion

of environmentally conscious business practices among suppliers is fostered by greater relational

interdependence (Tate et al., 2013). Managing sustainability in BSRs is essentially a reciprocal

concern (Chen et al., 2017), and the adoption and implementation of sustainability practices are

particularly active when buyers can exert coercive power (Sancha et al., 2019; Touboulic et al.,

11

2014). Thus, it is also conceivable that suppliers might adhere to the sustainability standards

set by the buying firm to preserve relationships and secure business opportunities with them. At

the same time, the level of supplier dependence determines the buying firms’ ability to monitor

and encourage cooperation to facilitate each other and improve their environmental performance

(Hajmohammad et al., 2024; Kim and Henderson, 2015; Kim and Wemmerlöv, 2015; Hoejmose

et al., 2013; Gimenez and Sierra, 2013). Therefore, the greater the supplier’s dependence on

the buyer, the greater the benefits the supplier obtains by implementing collaborative practices to

enhance the sustainability of their operations (Sancha et al., 2019; Touboulic et al., 2014). To

summarize, supplier dependence has been considered in the literature as a key contingency factor

that influences the effectiveness of sustainability practices among suppliers. Thus, we hypothesize

the following:

Hypothesis 3A (H3A). The positive relationship between the environmental efficiency of the

first-tier suppliers and that of the focal firms is strengthened by the financial dependence of first-tier

suppliers on focal firms.

Hypothesis 3B (H3B). The positive relationship between the environmental efficiency of the

second-tier suppliers and that of the first-tier suppliers is strengthened by the financial dependence

of second-tier suppliers on first-tier suppliers.

2.3 Research Design

In this section, we first discuss the structure of MSCs used in our analysis. Next, we explain

our data collection procedure and the sources of our data. We then develop our NDEA model

to compute comprehensive measures of environmental efficiency. Finally, we define the variables

used to test our hypotheses.

2.3.1 Triads

Different structural arrangements imply different business connections and outcomes between

member firms in supply chains. In the context of MSCs, structural arrangements involve different

combinations of member firms. As such, triads, the fundamental elements of supply chains, can be

formed in various ways (Choi and Wu, 2009). Based on the structural, contextual, and relational

12

characteristics, previous work has predominantly presented two major forms of triads: open and

closed triadic structures (Vedel et al., 2016; Tachizawa and Wong, 2014; Mena et al., 2013).

Specifically, groups of three interconnected firms create either an open triad where the three firms

are indirectly linked through one of them, as shown in Figure 2.1, or a closed triad where all firms

are directly linked to each other (Vedel et al., 2016).

Figure 2.1 Open Triad

Supplier’s
Supplier

Supplier

Buyer

Based on diverse methods, such as surveys, case studies, and conceptual and statistical analyses,

previous studies have used different forms of triadic relationships (Vedel et al., 2016; Wu et al., 2010;

McFarland et al., 2008). Drawing upon field interviews and institutional theory, the propagation

of interfirm behaviors using two adjacent dyads (that is, manufacturer, dealer, and customer triads)

is examined (McFarland et al., 2008), while supplier-supplier co-opetition using triadic data (that

is, buyer-supplier-supplier triads) is also empirically investigated (Ried et al., 2021; Wu et al.,

2010). Since our study aims to understand the cross-tier environmental impact in MSCs beyond

the first-tier suppliers, using open triadic relationships is more appropriate. Focusing on open

triadic relationships allows us to aggregate and compute tier-level metrics across the three tiers of

MSCs. We note that closed triadic relationships are not appropriate for our study because, in those

relationships, the supplier’s supplier is also a direct supplier of the focal firm since they can establish

a direct connection with each other. Thus, our study takes advantage of open triadic relationships

to isolate second-tier suppliers, which do not have a direct and contract-based relationship with

the focal firm, from focal companies (Jamalnia et al., 2023; Osadchiy et al., 2021; Wilhelm et al.,

2016a; Grimm et al., 2014). Further, considering open triadic relationships allows us to utilize the

NDEA model to compute tier-level environmental efficiency scores.

13

Figure 2.2 Data Collection and Processing

Step 1

1. Identify buyer-supplier relationships using revenue percentage data (FactSet).

a) Among the different types of normalized relationships captured by FactSet Revere, select customer

relationships with percentage revenues.

2. Merge buyer-supplier relationship data (FactSet) with environmental performance data (Trucost), spanning

2014 to 2018 (12,231 dyads), using International Securities Identification Number (ISIN).

Step 2

1. Remove duplicate buyer-supplier relationships (8,222 dyads).

a) Remove duplicate buyer-supplier relationships with the same unique identifier for the relationship

record in a year between 2014 and 2018 (8,675 dyads).

b) Remove duplicate reporting of 451 buyer-supplier relationships with the same buyer and supplier but

different relationship identifiers based on two criteria (8,224 dyads).

i. Remove redundant buyer-supplier relationships with different unique identifiers for the

relationship record but the same duration and percentage revenues.

ii. Remove redundant buyer-supplier relationships with relatively small percentage revenues.

c) Remove buyer-supplier relationships with the same start/end date and zero percentage revenue

(8,222 dyads).

Step 3

1. Create triadic relationships using the multiple dyadic relationships established in Step 3 (1,067 triads).

a) Remove triads with missing environmental performance data (1,911 triads).

b) Remove triads where a focal firm’s ISIN is equivalent to the second-tier supplier’s ISIN (1,851

triads).

c) Select focal firms operating in 10 four-digit Global Industry Classification Standard (GICS) groups,

all within the manufacturing category (1,202 triads).

d) Remove focal firms with less than $1 billion annual revenues (1,195 triads).

e) Remove triads that have direct business connections between second-tier suppliers and focal firms
(to sift out closed triadic relationships, we extensively examined 637,313 FactSet Revere supply
chain relationships data beyond 2014) (1,067 triads).

2. Aggregating the triadic relationships established in the previous step, create triads with unique focal firm

and year combinations (289 firm-level observations).

Step 4

1. Drop 30 observations due to NDEA empirical rules (259 firm-level observations).

14

2.3.2 Data

In this section, we discuss our data collection procedure and the sources of our data set. An

overview of our data collection and processing procedure is provided in Figure 2.2.

Our empirical analysis is based on multiple buyer-supplier-supplier triads in MSCs. To con-

struct these triadic relationships, we utilize firm-level BSR data provided by FactSet Revere. The

effectiveness of FactSet Revere’s supply chain relationships data has been highlighted within the

context of SCM (Osadchiy et al., 2021; Wang et al., 2021) and environmental management (Cole

et al., 2023; Modi and Cantor, 2021). According to FactSet Revere, their global business rela-

tionship data is hand-collected and verified from various public sources, including annual filings,

investor presentations, and press releases (Agca et al., 2022; Wu et al., 2022). The established BSRs

are also tracked daily based on public releases and corporate actions, ensuring that inter-company

relationships are up-to-date. Prior research using FactSet Revere supply chain relationships data

has utilized the start and end dates of BSRs to create variables and construct either time series

or panel data for their studies (Pankratz and Schiller, 2024; Ağca et al., 2023; Crosignani et al.,

2023; Culot et al., 2023; Gofman and Wu, 2022; Wu et al., 2022; Osadchiy et al., 2021; Wang

et al., 2021; Chae et al., 2020; Gofman et al., 2020; Schiller, 2018). In particular, Gofman and Wu

(2022) uses the start and end dates to create a snapshot of the BSRs observed at the end of each

calendar year, enabling the formation of a series of production networks at an annual frequency.

Crosignani et al. (2023) also discusses the superiority and granularity of FactSet Revere supply

chain relationships data, mentioning that it offers an adequately frequent occurrence of the start

and end dates of supply chain relationships. Culot et al. (2023) specifically mentions that start

dates of BSRs are determined by the date when the information on BSRs is processed, whereas end

dates are decided by the announcement of source companies and verified over time. In this study,

we use this unique feature of FactSet Revere to set up a panel data structure comprising multiple

buyer-supplier-supplier triads for our empirical analysis.

The second distinctive feature of FactSet Revere supply chain relationships data is that it

provides information on revenue dependence in BSRs. According to FactSet Revere, their revenue

15

percent data indicates the percentage of revenue a supplier derives from their relationship with each

downstream member between the start and end dates. We use this revenue percent data not only to

test the moderating effect of supplier revenue dependence on cross-tier environmental impact but

also to compute the environmental efficiency scores of the member firms in all the triads in our

data using our NDEA model (Pankratz and Schiller, 2024; Hyun and Kim, 2018; Schiller, 2018).

Another data set used in this study is Trucost environmental data (De Stefano and Montes-

Sancho, 2024; Cole et al., 2023). Trucost provides data on hundreds of key environmental perfor-

mance indicators across 500 distinct industry sectors via a four-step research process: mapping,

estimating, collecting, and engaging with companies. This data set also relies on diverse sources,

such as direct corporate disclosure of more than 15,000 corporations, scientific literature, and

global-level and national-level databases, to collect a variety of information on pollutants, wa-

ter dependency, natural resource consumption, waste disposal, corporate disclosure, and annual

revenue. Based on prior research in SSCM (as shown in Table 2.1), to compute environmental

efficiency scores, we collect data from the Trucost database on four different environmental per-

formance indicators, in addition to revenue data, for each member of the triadic relationships we

develop using FactSet data.

In our analysis, to control for heterogeneity in our hypotheses tests, we include several vari-

ables in our econometrics models, such as industry similarity and geographical distance between

neighboring partners. We also include the economic freedom score of the country to which the

firm under study belongs. Prior research has considered economic freedom as a driver of social

and environmental responsibility (Graafland, 2019; Hartmann and Uhlenbruck, 2015), and hence,

we include the economic freedom data from the Heritage Foundation as a control variable in our

analysis. This data provides economic freedom scores for a country based on the rule of law,

government size, regulatory efficiency, and market openness. We obtain the geographical distance

data from the open-source website Geodatos1.

Using the FactSet and Trucost datasets, we first identify BSRs with revenue percentage data

1https://www.geodatos.net/en

16

from 2014 to 2018. Next, we utilize these multiple dyadic relationships to form various triadic

relationships. Finally, as in prior research (Song et al., 2023; Barker et al., 2022; Osadchiy et al.,

2021; Dong et al., 2020), we aggregate these triadic relationships to create triads with unique

focal firm and year combinations, resulting in 259 firm-level observations spanning from 2014 to

2018. The details of this procedure are explained step-by-step in Figure 2.2. In our dataset, for the

observation of a focal firm’s triad in a given year, we compute every variable related to tier-level

performance or characteristic by appropriately aggregating or averaging the metric across different

firms in that tier. We explain the details in the next sections. Our final dataset with 259 observations

is an unbalanced panel with observations for each focal firm ranging from 2014 to 2018. Note

that some focal firms may lack observations in specific years due to the unavailability of revenue

percentage data for the dyadic relationships within their triads during those years.

2.3.3 Computing Environmental Efficiency Scores using NDEA

In this study, we use the data envelopment analysis (DEA) technique to compute the environ-

mental efficiency scores of firms at different tiers of MSCs. DEA is a method used for assessing

the relative efficiency of peer decision-making units (DMUs), and this technique is frequently used

in SCM research to analyze and compare performance across business units. Efficiency scores

obtained through DEA methods are commonly used as variables in subsequent empirical studies,

thereby extending the applicability and insights provided by DEA (Dreyfus et al., 2020; Kao et al.,

2017; Jacobs et al., 2016).

DEA requires an assumption of a homogeneous set of units. Therefore, we categorize our DMUs

according to the Global Industry Classification Standard (GICS) industry taxonomy to enable a

comparison between homogeneous DMUs. Furthermore, for our analysis in this study, we consider

the aggregated triadic relationship for each focal firm within a given year. To reflect this, we treat

the (aggregated) triad2 of a focal company as a DMU (Tavana et al., 2013; Chen and Yan, 2011);

Figure 2.3 illustrates the DMU of our analysis, which is a triad comprising three different tiers:

focal firm, first-tier suppliers, and second-tier suppliers. A limitation of the DEA method is that it

2Henceforth, whenever we mention the triad of focal firm 𝑖 in year 𝑡, we are referring to the aggregated triad of

focal firm 𝑖 in year 𝑡.

17

treats each DMU as a black box, relying solely on the initial inputs and final outputs consumed and

produced by the DMU, respectively, and does not capture the internal and linking activities between

divisions of DMUs. For our study, including the linking activities of DMUs is essential to capture

the network interrelationships between member firms. Hence, to overcome the drawback of DEA

and to capture the internal linking activities between divisions of DMUs, we use a NDEA model by

introducing the concept of intermediate variables that link two consecutive divisions so that parts

of the outputs from the preceding division can be used as inputs to the subsequent division (Tone

and Tsutsui, 2009).

Figure 2.3 NDEA Configuration

Since industrial production and business operations require diverse input resources and generate

multiple negative environmental outcomes, we develop an NDEA model that takes into account

multiple desirable and undesirable environmental outcomes to compute environmental efficiency

scores for members of MSCs. Table 2.1 presents the input and output factors used in our NDEA

model. We now discuss how we operationalize the input, output, and intermediate factors at each

tier of the triadic DMU. First, since the triadic relationships we study comprise only one focal firm,

we directly use the focal firm’s inputs and outputs as the corresponding inputs and outputs in the

NDEA model. Let 𝐼𝑛𝑝𝑢𝑡𝑖𝑡 (resp., 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖𝑡) be an input (resp., output) of a focal firm 𝑖 in year 𝑡.

Then, in our NDEA model, for the triad formed by focal firm 𝑖 in year 𝑡, we define the input and

18

Name

Water Purchased

Variable (Label)
Input (𝑋1)

Natural Resource Use

Input (𝑋2)

Carbon-Scope 1 & 2 Emis-
sions

Undesirable Output (𝑌1)

Waste

Undesirable Output (𝑌2)

Table 2.1 NDEA Variables

Description

Reference

The volume of water purchased
from utility companies (cubic
meter)

External cost of direct and in-
direct natural
resource usage
(USD mn)

the greenhouse gas
Sum of
(GHG) emissions from opera-
tions that are owned or con-
trolled by the company and the
GHG emissions from consump-
tion of purchased electricity,
heat or steam by the company
(tonnes CO2e)

External cost of hazardous and
non-hazardous waste produced
by the company (USD mn)

Cole et al. (2023); Fu
and Jacobs (2022);
Sodhi
Tang
and
(2019)

Kalaitzi et al. (2018);
Brandenburg et al.
(2014); Tang
and
Zhou (2012)

(2024,
Song et al.
2023);
Adhikary
et al. (2020); Saun-
(2020);
ders et al.
Tang
and
Sodhi
(2019); Zhou et al.
(2018); Wu et al.
(2016)

(2016);
Wu et al.
Tang
Zhou
and
(2012); Kleindorfer
et al. (2005)

An et al. (2016); Wu
et al. (2016); Bran-
denburg et al. (2014)

Revenue

Intermediate (𝑍1)

Annual revenue (USD mn)

output of focal firm 𝑖 as follows:

𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐼𝑛𝑝𝑢𝑡𝑖𝑡 = 𝐼𝑛𝑝𝑢𝑡𝑖𝑡

𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖𝑡 = 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖𝑡

(2.1)

(2.2)

Next, in year 𝑡, let 𝐼𝑛𝑝𝑢𝑡𝑖 𝑗𝑡 (resp., 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖 𝑗𝑡) be an input (resp., output) of a first-tier supplier

𝑗 in the triad of firm 𝑖. Let 𝐽 be the total number of first-tier suppliers of focal firm 𝑖 in the triad.

Then, to consider the impact of network interrelationships between the focal firm and its first-tier

suppliers on cross-tier environmental influence, we aggregate the input and output of first-tier

suppliers within the triad of focal firm 𝑖 as follows:

19

𝑇𝑖𝑒𝑟1 𝐼𝑛𝑝𝑢𝑡𝑖𝑡 =

𝑇𝑖𝑒𝑟1 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖𝑡 =

𝐽
∑︁

𝑗=1
𝐽
∑︁

𝑗=1

𝐼𝑛𝑝𝑢𝑡𝑖 𝑗𝑡

𝑂𝑢𝑡 𝑝𝑢𝑡𝑖 𝑗𝑡

(2.3)

(2.4)

Moreover, we capture the network interrelationships between the focal firm and its first-tier suppliers

using the total amount of revenue of first-tier suppliers that is dependent on focal firm 𝑖 as the linking

variable (variable 𝑍1 as shown in Figure 2.3). In year 𝑡, for the triad of focal firm 𝑖, using the revenue

of first-tier firm 𝑗 (𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗𝑡), the percentage of revenue dependence of first-tier supplier 𝑗 on

focal firm 𝑖 (𝑅𝑃𝑖 𝑗𝑡), the relationship duration (in days) between first-tier supplier 𝑗 and focal firm

𝑖 (𝑅𝐷𝑖 𝑗𝑡), and the number of days in that year (𝑁𝑡), we define the linking variable connecting the

first tier of the triad to the focal firm as follows:

𝑇𝑖𝑒𝑟1 𝐼𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑖𝑡 =

𝐽
∑︁

𝑗=1

𝑅𝐷𝑖 𝑗𝑡
𝑁𝑡

× 𝑅𝑃𝑖 𝑗𝑡 × 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗𝑡

(2.5)

Note that since the focal firm 𝑖 and first-tier supplier 𝑗 may not always maintain a business

relationship throughout the entire year 𝑡, we compute our linking variable based on the fraction

of the year 𝑡 during which their business relationship exists. Finally, to capture the network

interrelationships between the second-tier suppliers and their downstream firms, we use a similar

approach to the one discussed above. We define the input, output, and linking variables for the

second tier of the triad of focal firm 𝑖 as follows:
𝐾
∑︁

𝑇𝑖𝑒𝑟2 𝐼𝑛𝑝𝑢𝑡𝑖𝑡 =

𝐽
∑︁

𝐼𝑛𝑝𝑢𝑡𝑖 𝑗 𝑘𝑡

𝑇𝑖𝑒𝑟2 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖𝑡 =

𝑇𝑖𝑒𝑟2 𝐼𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝑀𝑒𝑎𝑠𝑢𝑟𝑒𝑖𝑡 =

𝑘=1
𝐾
∑︁

𝑗=1
𝐽
∑︁

𝑘=1
𝐾
∑︁

𝑗=1
𝐽
∑︁

𝑂𝑢𝑡 𝑝𝑢𝑡𝑖 𝑗 𝑘𝑡

𝑅𝐷𝑖 𝑗 𝑘𝑡
𝑁𝑡

× 𝑅𝑃𝑖 𝑗 𝑘𝑡 × 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗 𝑘𝑡

(2.8)

𝑘=1
Here, in year 𝑡, 𝐼𝑛𝑝𝑢𝑡𝑖 𝑗 𝑘𝑡 (resp., 𝑂𝑢𝑡 𝑝𝑢𝑡𝑖 𝑗 𝑘𝑡) denotes an input (resp., output) of a second-tier

𝑗=1

supplier 𝑘 (which is the supplier of first-tier supplier 𝑗) in the triad of focal firm 𝑖. Further, in year

20

(2.6)

(2.7)

𝑡, the variables 𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗 𝑘𝑡, 𝑅𝑃𝑖 𝑗 𝑘𝑡, and 𝑅𝐷𝑖 𝑗 𝑘𝑡 represent the revenue of the second-tier supplier

𝑘, the percentage of revenue dependence of second-tier supplier 𝑘 on first-tier supplier 𝑗, and the

relationship duration (in days) between second-tier supplier 𝑘 and first-tier supplier 𝑗, respectively.

Using all the inputs, outputs, and linking variables defined above for each tier of the triadic

relationships, we run our NDEA model to compute the environmental efficiency scores for each

tier. Based on the homogeneity assumption of NDEA, we run separate NDEA models by clustering

all the triads of focal firms within the same industry, according to the GICS classification. In each

NDEA model we run for a cluster of triads belonging to the same industry, we treat the triad of a

focal firm in a given year as a separate DMU (Wang, 2019; Surroca et al., 2016). This procedure

benchmarks all DMUs, enabling meaningful comparisons of their performance through efficiency

scores (Wang, 2019). Further details of our NDEA model can be found in the Appendix.

For the triad of each focal firm 𝑖 in a given year 𝑡, our NDEA models compute three environmental

efficiency scores, one for each tier. We represent these as follows: 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚𝑖𝑡, 𝑇𝑖𝑒𝑟1 𝐸 𝐸𝑖𝑡,

𝑇𝑖𝑒𝑟2 𝐸 𝐸𝑖𝑡, corresponding to the environmental efficiency score of focal firm 𝑖, the first-tier of focal

firm 𝑖, and the second-tier of focal firm 𝑖, respectively, in year 𝑡.

2.3.4 Variables for Econometric Analysis

In this section, we first describe the dependent and independent variables, followed by the

control variables we use in our econometric models.

2.3.4.1 Dependent Variables

To test the hypotheses of our study, we use two dependent variables: environmental efficiency

scores of first- and second-tier suppliers. As discussed in the previous section, using NDEA analysis,

we first compute the dependent variable 𝑇𝑖𝑒𝑟2 𝐸 𝐸𝑖𝑡. This variable is a score that collectively

measures the environmental efficiencies of all the second-tier suppliers in the triad of focal firm

𝑖 in year 𝑡. Note that this score is based on multiple input, output, and intermediate measures

in the upstream stage of Figure 2.3, and comprehensively mirrors the environmental footprint of

second-tier suppliers. Similarly, we compute our second dependent variable 𝑇𝑖𝑒𝑟1 𝐸 𝐸𝑖𝑡, a score

that collectively measures the environmental efficiencies of all the first-tier suppliers in the triad of

21

focal firm 𝑖 in year 𝑡.

2.3.4.2

Independent Variables

Testing our hypotheses requires four main independent variables. To test the influence of the

first-tier suppliers on the environmental efficiency of the second-tier suppliers, we use 𝑇𝑖𝑒𝑟1 𝐸 𝐸𝑖𝑡

as the independent variable. Next, to test the influence of the focal firm on the environmental

efficiency of the first-tier suppliers, we compute the independent variable 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸𝑖𝑡 using

our NDEA model.

To test H3A and H3B, we compute the following two independent variables: 𝑇𝑖𝑒𝑟1𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡

and 𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡. The variable 𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡 (resp., 𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡) measures

the degree of first-tier (resp., second-tier) suppliers’ revenue dependence on the focal firm (resp.,

first-tier suppliers). Below, we describe the procedure used to compute these variables.

Consistent with prior studies (Pankratz and Schiller, 2024; Hyun and Kim, 2018; Schiller, 2018),

using FactSet Revere, we obtain supplier revenue dependence data, which measures the percentage

of a supplier’s revenue derived from the relationship with their downstream partner. Consider focal

firm 𝑖 in year 𝑡. Let 𝑅𝑃𝑖 𝑗𝑡 be the percentage revenue dependence of first-tier supplier 𝑗 on the focal

firm 𝑖 in year 𝑡. Then, we compute 𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡 – the overall revenue dependence of its

first-tier suppliers on focal firm 𝑖 in year 𝑡 – as a weighted average of the revenue dependencies of

all its first-tier suppliers as follows:

𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡 =

𝐽
∑︁

(cid:16)

𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗𝑡

(cid:205)𝐽

𝑗=1

𝑗=1

𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗𝑡

(cid:17)

×

𝑅𝐷𝑖 𝑗𝑡
𝑁𝑡

× 𝑅𝑃𝑖 𝑗𝑡 .

(2.9)

The variables in the above equation are defined as in (2.5). Note that since a first-tier supplier does

not always have a relationship with the focal firm throughout the year, our method uses the exact

proportions of the relationship durations as weights to compute the average revenue dependence.

Using a similar approach, we compute the overall revenue dependence of the second-tier suppliers

(of firm 𝑖) on the first-tier suppliers (of firm 𝑖) as follows:

𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡 =

𝐽
∑︁

(cid:16)

𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗𝑡

(cid:205)𝐽

𝑗=1

𝑗=1

𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗𝑡

(cid:17)

×

𝐾
∑︁

(cid:16)

(cid:16)

𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗 𝑘𝑡

(cid:205)𝐾

𝑘=1

𝑘=1

𝑅𝑒𝑣𝑒𝑛𝑢𝑒𝑖 𝑗 𝑘𝑡

(cid:17)

×

𝑅𝐷𝑖 𝑗 𝑘𝑡
𝑁𝑡

× 𝑅𝑃𝑖 𝑗 𝑘𝑡

(cid:17)

(2.10)

22

The variables in the above equation are defined as in (2.8).

2.3.4.3 Control Variables

Several firm-level and country-level characteristics across different tiers in MSCs may influence

the environmental efficiency scores of the member firms. To account for potential confounding

effects and the impact of unobserved factors, we include a set of control variables in our empirical

models. We discuss this below in detail.

Previous work shows that a firm’s financial performance measure is associated with corporate

willingness and ability to disclose environmental performance (Bellamy et al., 2020; Jira and Toffel,

2013; Al-Tuwaijri et al., 2004). Furthermore, a focal firm’s size is often seen as an influential factor

of environmental performance (Modi and Cantor, 2021) and used as a measure for a buyer’s power

(Jira and Toffel, 2013). Therefore, to account for a focal firm’s size and power in its supply chains,

we use its annual revenue as a control variable. In our models, we include the control variable

lnRevenue𝑖𝑡, which is the natural logarithm of the annual revenue of a focal firm 𝑖 in year 𝑡.

Since partnership success is contingent on sharing similarities in corporate strategies, struc-

ture, and cultural fit, companies take advantage of same-sector partnerships to increase efficiency

(Gutiérrez et al., 2016). Previous work on managing corporate sustainability in the context of MSCs

controls for information on suppliers’ industry heterogeneity (Song et al., 2024, 2023; Bellamy et al.,

2020; Jira and Toffel, 2013). Therefore, we include another set of control variables in our models

that represent whether two firms in a dyad are included in the same industry. For the triad of focal

firm 𝑖 in year 𝑡, using Trucost data, we compute the control variable Same Industry Downstream𝑖𝑡

as the proportion of first-tier firms belonging to the same industry as the focal firm 𝑖. Similarly, for

the triad of focal firm 𝑖 in year 𝑡, we compute the control variable Same Industry Upstream𝑖𝑡 as the

average of the variables Same Industry Upstream𝑖 𝑗𝑡, where Same Industry Upstream𝑖 𝑗𝑡 represents

the proportion of second-tier firms of first-tier firm 𝑗 belonging to the same industry as the first-tier

firm 𝑗.

Geographical location and dispersion of partner firms are other decisive factors that determine

the implementation of sustainability practices (Wilhelm et al., 2016b) and increase supply chain

23

complexity (Wilhelm et al., 2016a; Awaysheh and Klassen, 2010; Bozarth et al., 2009). Hence,

we also control for the average between-firm geographical distances across different tiers of supply

networks. The control variable lnUpstream Distance𝑖𝑡 (resp., lnDownstream Distance𝑖𝑡) in our

models represents the natural logarithm of the average geographical distances between second-tier

suppliers and first-tier suppliers (resp., between first-tier suppliers and focal firm) of focal firm 𝑖 in

year 𝑡. The procedures used to compute these variables are detailed below.

Across multiple triadic relationships, we gather sets of cross-country distance data, comprising

the distance between centroids of the two countries. Using Geodatos, these measurements are taken

between the two countries where the headquarters of the two firms, engaged in a dyadic relationship

within supply networks, are situated. For a triad formed with firm 𝑖 as the focal firm, in year 𝑡,

let 𝐷𝑖 𝑗𝑡 denote the distance between the centroids of the countries in which the focal firm 𝑖 and its

first-tier firm 𝑗 are located. Then, to control for the effect of the geographical distance between

focal firm 𝑖 and its first-tier firms on their cross-tier environmental impact, we define the variable

as follows:

𝑙𝑛𝐷𝑜𝑤𝑛𝑠𝑡𝑟𝑒𝑎𝑚 𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒𝑖𝑡 = ln (cid:169)
(cid:173)
(cid:171)

1
𝐽

𝐽
∑︁

𝑗=1

𝐷𝑖 𝑗𝑡 + 1(cid:170)
(cid:174)
(cid:172)

(2.11)

Note that since distances between countries in our data follow a skewed distribution, we apply a

natural logarithmic transformation to normalize the distances. Further, we add 1 to the average

geographical distances before applying the logarithm to avoid any negative values. Similarly, to

control for the effect of the geographical distance between the second-tier and first-tier of focal firm

𝑖 on their cross-tier environmental impact, we define the variable as follows:

𝑙𝑛𝑈 𝑝𝑠𝑡𝑟𝑒𝑎𝑚 𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒𝑖𝑡 = ln (cid:169)
(cid:173)
(cid:171)

1
𝐽

𝐽
∑︁

𝑗=1

1
𝐾

𝐾
∑︁

𝑘=1

𝐷𝑖 𝑗 𝑘𝑡 + 1(cid:170)
(cid:174)
(cid:172)

(2.12)

Here, 𝐷𝑖 𝑗 𝑘𝑡 represents the distance between the centroids of the countries in which the second-tier

firm 𝑘 and the first-tier firm 𝑗 (of the focal firm 𝑖) are located in year 𝑡.

Across various nations, economic freedom has been demonstrated to enhance the ability to

innovate in environmentally sustainable ways. Significant advancements in clean energy utilization

24

Table 2.2 Descriptive Statistics and Correlations

Mean SD

Min Max 1

N = 259
1. 𝑇𝑖𝑒𝑟2 EE
1.00
0.00
0.296
0.164
2. 𝑇𝑖𝑒𝑟1 EE
1.00
0.00
0.373
0.304
3. Focal Firm EE
1.00
0.348
0.03
0.678
4. 𝑇𝑖𝑒𝑟2 Dependence (%)
100
15.921 0.13
9.367
5. 𝑇𝑖𝑒𝑟1 Dependence (%)
83.17 0.04
8.386
10.839 0.00
6. Same Industry Upstream (%)
100
47.881 39.917 0.00
7. Same Industry Downstream (%) 46.416 47.593 0.00
100
8. lnUpstream Distance
9.41
0.00
9. lnDownstream Distance
0.00
9.45
10. Economic Freedom
52.00 90.20 0.08
11. lnRevenue
7.54

4.000
5.189
4.472
4.294
71.653 7.696
10.468 1.238

2

3

4

5

6

7

1.00
0.52* 1.00
0.14* 0.46* 1.00
-0.05 -0.12

-0.04 1.00

-0.12* 1.00
0.21* 0.10
0.25* -0.20* 1.00
-0.10 -0.13* 0.08
0.06
0.12
0.02
-0.00 -0.11
1.00
0.03
0.14* -0.01
0.10
0.01
0.18* -0.07
0.07
0.11
-0.01
-0.01
0.14* 0.24* 0.23* 0.12
-0.05
0.01
0.06
0.11
0.11
0.29* -0.16* 0.06
0.23* -0.04

0.10
12.54 -0.08 0.07

Mean SD

Min Max 8

9

10

11

8. lnUpstream Distance
9. lnDownstream Distance
10. Economic Freedom
11. lnRevenue

Note: ∗𝑝 < 0.05.

4.000
5.189
4.472
4.294
71.653 7.696
10.468 1.238

9.41
9.45

1.00
0.00
0.00
0.11
52.00 90.20 0.32* 0.29* 1.00
7.54

1.00

12.54 0.28* 0.38* 0.28* 1.00

and energy efficiency have predominantly arisen from improvements in economic freedom and

expanded trade (Miller et al., 2022). Economic freedom is measured by multiple quantitative and

qualitative factors in four broad categories, including rule of law, government size, regulatory

efficiency, and open markets, and it is known to stimulate corporate environmental responsibility

(Graafland, 2019). Thus, in our analysis, we control for the overall economic freedom of the country

in which the focal firm is headquartered. We denote this control variable by Economic Freedom𝑖𝑡.

In summary, the final data set for our empirical analysis is panel data, ranging from 2014 to

2018. The descriptive statistics of all the variables in our study and their correlations are presented

in Table 2.2.

2.4 Analysis and Results

In this section, we first discuss the results of our NDEA model. Next, we describe the models

used for the empirical analysis to test our hypotheses. Finally, we present the results of our empirical

models.

25

2.4.1 NDEA Results

Recall that for the triad of each focal firm in years ranging from 2014 to 2018 (i.e., for all the

259 unique focal firms and year combinations in our panel dataset), our NDEA model outputs three

efficiency scores: one each for the focal firm, the first tier, and the second tier. Table 2.3 provides

the averages and standard deviations of the environmental efficiency scores of focal firms, first-tier

suppliers, and second-tier suppliers. As shown in the table, all focal firms in our dataset fall under

the general category of manufacturing and are classified into six distinct GICS industry groups.

Additionally, the first and second-tier suppliers belong to 15 different industry groups, each with

its own industry composition, as detailed in Table 2.4.

Table 2.3 Sample Statistics of Focal Firms across 6 GICS Industries

𝑻𝒊𝒆𝒓2 𝑬 𝑬

𝑻𝒊𝒆𝒓1 𝑬 𝑬

𝑭𝒐𝒄𝒂𝒍 𝑭𝒊𝒓𝒎 𝑬 𝑬

Industry (four-digit GICS code)

Observations Mean Std. dev. Mean Std. dev. Mean Std. dev.

Automobiles & Components (2510)
Capital Goods (2010)
Health Care Equipment (3510)
Materials (1510)
Pharmaceuticals & Biotechnology (3520)
Technology Hardware & Equipment (4520)

Total

54
60
31
55
27
32

259

0.114
0.164
0.161
0.093
0.353
0.210

0.164

0.243
0.306
0.275
0.263
0.316
0.353

0.296

0.283
0.178
0.474
0.339
0.260
0.385

0.304

0.346
0.334
0.362
0.402
0.369
0.381

0.373

0.747
0.451
0.843
0.666
0.772
0.772

0.678

0.273
0.388
0.208
0.387
0.240
0.306

0.348

Table 2.4 Industry Composition of First- and Second-tier Suppliers

Second-tier Suppliers

First-tier Suppliers

GICS Industry Classifications

Observations GICS Industry Classifications

Observations

Automobiles & Components
Capital Goods
Commercial & Professional Services
Consumer Durables & Apparel
Energy
Food, Beverage & Tobacco
Health Care Equipment
Materials
Pharmaceuticals & Biotechnology
Real Estate
Semiconductors & Semiconductor Equipment
Software & Services
Technology Hardware & Equipment
Transportation
Utilities

114
59
2
1
63
1
3
99
275
254
83
23
19
3
18

Automobiles & Components
Capital Goods
Consumer Durables & Apparel
Energy
Food & Staples Retailing
Food, Beverage & Tobacco
Health Care Equipment
Materials
Pharmaceuticals & Biotechnology
Real Estate
Semiconductors & Semiconductor Equipment
Software & Services
Technology Hardware & Equipment
Transportation
Utilities

Note: Observations are reported based on the number of unique firms in a given year.

97
31
5
37
3
5
6
64
153
3
27
2
17
4
6

26

2.4.2 Empirical Strategy

We now discuss the empirical models we use to test our hypotheses. To test the influence of the

focal firm’s environmental efficiency on that of its first-tier suppliers (H1A) and if this influence is

affected by the revenue dependence of the first-tier suppliers, we use the following model:

𝑇𝑖𝑒𝑟1 𝐸 𝐸𝑖𝑡 = 𝛾0 + 𝛾1 × 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸𝑖𝑡 + 𝛾2 × 𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡

+ 𝛾3 × 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸𝑖𝑡 × 𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡 + 𝜸4 × 𝜼𝒊𝒕 + 𝛼𝑖 + 𝛿𝑡 + 𝜖𝑖𝑡 .

(2.13)

In this model, 𝜼𝒊𝒕 represents all the control variables such as revenue and the economic freedom

of the focal firm (discussed in Section 2.3.4.3), whereas the variables 𝛼𝑖 and 𝛿𝑡 control for the focal

firm and year specific unobserved heterogeneity, respectively.

Next, to test the influence of the first-tier suppliers and focal firm on the environmental efficiency

of the second-tier suppliers (H1B), and if this influence is affected by the revenue dependence of

the second-tier suppliers, we use the following model:

𝑇𝑖𝑒𝑟2 𝐸 𝐸𝑖𝑡 = 𝛽0 + 𝛽1 × 𝑇𝑖𝑒𝑟1 𝐸 𝐸𝑖𝑡 + 𝛽2 × 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸𝑖𝑡 + 𝛽3 × 𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡

+ 𝛽4 × 𝑇𝑖𝑒𝑟1 𝐸 𝐸𝑖𝑡 × 𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒𝑖𝑡 + 𝜷5 × 𝝁𝒊𝒕 + 𝛼𝑖 + 𝛿𝑡 + 𝜖𝑖𝑡 .

(2.14)

Here, 𝝁𝒊𝒕 denotes the set of control variables, while 𝛼𝑖 and 𝛿𝑡 account for unobserved hetero-

geneity at the focal firm and year levels, respectively.

Note that the efficiency score outputs from DEA models are typically numbers between 0 and

1. Therefore, for hypotheses tests on efficiency scores, we use a correlated random effects (CRE)

approach via the use of Mundlak device (Mundlak, 1978) to estimate nonlinear and unbalanced

panel data models in the presence of unobserved heterogeneity (Bates et al., 2024; Joshi and

Wooldridge, 2019; Wooldridge, 2019). The fraction probit model we use is as follows:

𝐸 [𝑌𝑖𝑡 |𝑋𝑖𝑡, ¯𝑋𝑖, 𝑇𝑖] = Φ(𝜓𝑡 + 𝑋𝑖𝑡 𝛽 + 𝑇𝑖𝛾 + ¯𝑋𝑖𝛿)

(2.15)

In the equation above, 𝑌𝑖𝑡 denotes fractional dependent variables, whereas 𝑋𝑖𝑡 collectively represents

independent variables, control variables, and year dummies (as described in (2.13) and (2.14)). The

27

variable ¯𝑋𝑖 = 𝑇 −1

𝑖

(cid:205)𝑇𝑖
𝑡=1

𝑋𝑖𝑡 refers to the time averages of each covariate, the parameter 𝜓𝑡 represents

year-specific intercepts, and Φ(·) denotes the cumulative distribution function (CDF) of the standard

normal distribution. Here, the time averages of explanatory variables serve as proxies for firm-level

fixed effects. To account for our unbalanced panel data, similar to that in Bates et al. (2024), we

employ indicators (𝑇𝑖) that denote the number of time periods observed for a focal firm 𝑖. In the

next section, we discuss the parameters estimated from these models and the corresponding results.

2.4.3 Main Results

Table 2.5 illustrates the results for all our hypotheses tests. The results for hypotheses H1A and

H3A are presented in Models 1, 2, and 3 in Panel A, while the results for hypotheses H1B, H2, and

H3B are presented in Models 4, 5, 6, and 7 in Panel B.

In the results of Models 2 and 3, the parameter estimate of 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸 is statistically

significant and positive supporting hypothesis H1A and indicating that the environmental efficiency

of the focal firm positively influences the efficiency of its first-tier suppliers. The parameter estimate

of 𝑇𝑖𝑒𝑟1 𝐸 𝐸 across Models 5, 6, and 7 is statistically significant and positive supporting hypothesis

H1B showing that the environmental efficiency of the first-tier suppliers positively influences the

efficiency of the second-tier suppliers. However, the estimate of 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸 in Models 6 and 7

is insignificant suggesting that focal firms do not directly influence the environmental efficiency of

second-tier suppliers, but rather do so indirectly through first-tier suppliers. Thus, our hypothesis

H2 is not supported.

Next, we examine the results of the hypothesis tests for H3A and H3B regarding the effect of

supplier revenue dependence on environmental efficiencies. In Model 3, the parameter estimate

of the interaction between 𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 and 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸 is statistically significant and

positive indicating that the revenue dependence of first-tier suppliers on the focal firm increases

the influence of the focal firm on their environmental efficiency. Specifically, at the mean level of

𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒, the marginal effect of 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸 on 𝑇𝑖𝑒𝑟1 𝐸 𝐸 is 0.502 (𝑠.𝑒. = 0.095),

and it is positive and statistically significant at the 0.1% level.

Similarly, the parameter estimate of the interaction between 𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 and 𝑇𝑖𝑒𝑟1 𝐸 𝐸 in

28

Table 2.5 Coefficient Estimates (APEs) of Pooled Fractional Probit Models

Dependent variable:

Independent variable:
𝑇𝑖𝑒𝑟1 𝐸 𝐸

𝑇𝑖𝑒𝑟1 𝐸 𝐸 (APE)

𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸

𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸 (APE)

𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒

𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 (APE)

𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 × 𝑇𝑖𝑒𝑟1 𝐸 𝐸

𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒

𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 (APE)

𝑇𝑖𝑒𝑟1 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒 × 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸

𝑆𝑎𝑚𝑒 𝐼𝑛𝑑𝑢𝑠𝑡𝑟 𝑦 𝑈 𝑝𝑠𝑡𝑟𝑒𝑎𝑚

𝑆𝑎𝑚𝑒 𝐼𝑛𝑑𝑢𝑠𝑡𝑟 𝑦 𝐷𝑜𝑤𝑛𝑠𝑡𝑟𝑒𝑎𝑚

𝑙𝑛𝑈 𝑝𝑠𝑡𝑟𝑒𝑎𝑚 𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒

𝑙𝑛𝐷𝑜𝑤𝑛𝑠𝑡𝑟𝑒𝑎𝑚 𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒

𝐸𝑐𝑜𝑛𝑜𝑚𝑖𝑐 𝐹𝑟𝑒𝑒𝑑𝑜𝑚

𝑙𝑛𝑅𝑒𝑣𝑒𝑛𝑢𝑒

Constant

Panel A

𝑻𝒊𝒆𝒓1 𝑬 𝑬

Panel B

𝑻𝒊𝒆𝒓2 𝑬 𝑬

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

1.618***
(0.257)
0.303***
(0.050)

1.687***
(0.271)
0.315***
(0.052)
-0.280
(0.265)
-0.052
(0.049)

1.387***
(0.340)
0.325***
(0.051)
-0.236
(0.258)
-0.043
(0.047)
-0.011
(0.009)
0.003
(0.002)
0.052*
(0.024)

1.867***
(0.478)
0.518***
(0.130)

-0.036
(0.458)
0.436***
(0.094)

-0.177***
(0.041)
0.004+
(0.002)
0.240***
(0.052)

-0.002
(0.004)

-0.002
(0.005)

-0.001
(0.004)

-0.039
(0.065)
-0.023+
(0.012)
-0.133
(0.321)
-1.495
(1.290)

-0.045
(0.070)
0.004
(0.018)
0.017
(0.361)
-1.313
(1.190)

-0.022
(0.063)
-0.002
(0.011)
-0.256
(0.250)
-0.614
(1.118)

-0.003
(0.006)
-0.004
(0.006)
-0.053
(0.035)
0.051
(0.035)
-0.005
(0.010)
-1.132**
(0.428)
-1.191
(1.539)

0.002
(0.005)
-0.004
(0.006)
-0.096**
(0.031)
0.090
(0.061)
0.010
(0.011)
-1.250***
(0.312)
-1.844
(1.755)

0.001
(0.005)
-0.004
(0.006)
-0.098**
(0.031)
0.089
(0.062)
0.005
(0.012)
-1.293***
(0.312)
-1.897
(1.710)

-0.000
(0.006)
-0.004
(0.005)
-0.095+
(0.054)
0.080
(0.073)
0.006
(0.012)
-1.097***
(0.321)
-1.663
(1.773)

Model Specification
Year Effects
Observations
Number of Focal Firms
Pseudo 𝑅2
Note: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical significance at the 10 percent, 5 percent, 1 percent, and 0.1
percent levels. Standard errors (𝑠.𝑒.) are clustered at the focal firm level and reported in parentheses below parameter estimates. We report
both the coefficients and average partial effects (APEs) from the fractional probit models. In addition, we present the pseudo 𝑅2 based on
squared correlation for all model specifications. The pooled fractional probit models include the time averages of independent variables, control
variables, and time dummies. For brevity, I do not report the APEs of control variables and the estimates of the time averages of independent
variables, control variables, and time dummies here.

Yes
259
115
0.184

Yes
259
115
0.464

Yes
259
115
0.204

Yes
259
115
0.386

Yes
259
115
0.420

Yes
259
115
0.351

Yes
259
115
0.399

Model 7 is statistically significant and positive indicating that the revenue dependence of second-tier

suppliers on first-tier suppliers increases the influence of the first-tier suppliers on their environ-

mental efficiency. To be specific, at the mean level of 𝑇𝑖𝑒𝑟2 𝐷𝑒 𝑝𝑒𝑛𝑑𝑒𝑛𝑐𝑒, the marginal effect of

29

𝑇𝑖𝑒𝑟1 𝐸 𝐸 on 𝑇𝑖𝑒𝑟2 𝐸 𝐸 is 0.348 (𝑠.𝑒. = 0.055), and it is positive and statistically significant at the

0.1% level. Consequently, both hypotheses H3A and H3B are supported.

Remark 1. (Note on Simultaneity Bias): We note that the parameter estimates of our models, and

consequently our results, are free from simultaneity bias. As discussed in Section 2.3.3, the inputs

and outputs used to compute the environmental efficiency for each tier differ from those used for

the other tiers not only, particularly due to differences in aggregation procedures and tier-specific

characteristics. Additionally, the linking variable in our NDEA models is revenue, which is not

a direct measure of environmental efficiency. Therefore, in any given triad, the environmental

efficiency score computed for first-tier suppliers (resp., second-tier suppliers) is not directly related

to the score computed for the focal firm (resp., first-tier suppliers), thereby eliminating concerns of

reverse causality and simultaneity bias.

2.5 Sensitivity Analysis

In this section, we explore how the influence of a downstream firm on its upstream partners

varies with the firm’s level of environmental efficiency. To this end, we build on the results from

the previous section, which employed a CRE framework using a fractional probit model.

The advantage of using fractional probit models with a nonlinear functional form is that it

enables the estimation of average partial effects (APEs) at different points along the distribution

of explanatory variables (Bates et al., 2024; Papke and Wooldridge, 2008). We run a pooled

fractional probit model with CRE estimation to estimate APEs at the 5th, 25th, 50th, 75th, and 95th

percentiles of both 𝑇𝑖𝑒𝑟1 𝐸 𝐸 and 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸 distributions. Using Models 2 and 5 in Table 2.5,

we examine how the influence of environmental efficiency changes at different percentiles of its

distribution.

Figures 2.4 and 2.5 illustrate how the environmental efficiency of downstream firms, at different

percentiles, affects the environmental efficiency of the upstream partners. Figure 2.4 shows the

influence of the focal firm’s environmental efficiency on first-tier suppliers, while Figure 2.5 shows

the influence of first-tier suppliers on second-tier suppliers. Both figures demonstrate a consistent

30

pattern of increasing influence at all percentiles of environmental efficiency. This suggests that at

any value of the environmental efficiency of the downstream firm, an increase in the efficiency will

continue to improve its impact on the upstream partner’s efficiency. Notably, its increasing rate is

the highest at intermediate levels of environmental efficiency in downstream firms. Additionally,

the impact of the focal firm’s environmental efficiency on first-tier suppliers is stronger compared to

the influence of first-tier suppliers on second-tier suppliers. This indicates that downstream supply

chain firms have a stronger influence on their upstream partners.

Figure 2.4 APEs of 𝐹𝑜𝑐𝑎𝑙 𝐹𝑖𝑟𝑚 𝐸 𝐸

Figure 2.5 APEs of 𝑇𝑖𝑒𝑟1 𝐸 𝐸

2.6 Discussion and Conclusion

2.6.1 Theoretical Implications

With increasing pressure to disclose firms’ supply chain information and the collective efforts of

supply chain members, significant progress has been made in reducing the social and environmental

impacts caused by supply chain participants (Bové and Swartz, 2016). However, to our knowledge,

previous studies seldom examine the dynamics of cross-tier environmental efficiency management

between firms at different supply chain levels. Furthermore, while existing studies on SSCM

predominantly focus on buying firms (Brandenburg et al., 2014), there is a growing need to assess

and monitor the sustainability practices of upstream suppliers in MSCs. Research on SSCM

that emphasizes the influence of downstream firms on upstream environmental outcomes remains

limited.

31

In this study, we aim to expand the scope of existing research by examining how the envi-

ronmental efficiency of downstream supply chain firms influences their upstream partners in an

open triadic configuration, where direct business connections between focal firms and second-tier

suppliers are nonexistent. We achieve this by aggregating firm-level environmental performance

data and utilizing NDEA to compute tier-level environmental efficiency scores for each triad. Our

empirical analyses reveal that downstream firms have a positive environmental influence on their

immediate upstream members.

Physical distance, measured by geographical, cultural, and organizational dimensions, under-

pins the rationale behind information asymmetry and coordination efforts (Tachizawa and Wong,

2014; Awaysheh and Klassen, 2010). This physical distance increases the likelihood of suppliers’

opportunistic behaviors and discretionary practices. Firms may adopt diverse social and environ-

mental practices, leading to different approaches to addressing environmental issues, which in turn

could provoke additional uncertainty due to limited access to information. Our empirical results in-

dicate that while focal firms positively influence the environmental efficiency of first-tier suppliers,

they do not have a significant impact on the environmental efficiency of second-tier suppliers. This

finding suggests that, in the absence of direct business connections, lower-tier suppliers are more

likely to align their environmental efficiency with their nearest partners rather than with entities

further down the supply chain.

We also examine the role of supplier revenue dependence in influencing the positive impact

of downstream firms on the environmental efficiency of upstream members. Power dynamics in

BSRs are inherently complex, with power asymmetry shaping the gains or losses from relational

dependence. However, the impact of dependency on the diffusion of environmentally friendly prac-

tices in MSCs remains under-explored (Tate et al., 2013). Therefore, we revisit power-dependence

relations from a resource-dependence and sustainability perspective to better understand how a

supplier’s revenue dependence influences environmental efficiency in a dyadic context. Our empir-

ical analysis reveals that supplier revenue dependence enhances the positive association between

the environmental efficiencies of dyads in supply chains. Our findings build upon prior research

32

on power-dependence relationships within SSCM, demonstrating that upstream partners’ environ-

mental efficiency improves with increased revenue dependence on downstream partners in BSRs.

2.6.2 Managerial Implications

The expansion of global supply chain networks and the growing financial and operational

interdependence among member firms have heightened interest among scholars and practitioners

in studying sustainability in SCM. Sustainability, represented by the confluence of the three main

pillars, people, planet, and profit, in operations management, encompasses green product and

process development, lean and green operations management, and remanufacturing and closed-

loop supply chains (Kleindorfer et al., 2005). However, the lack of sufficient knowledge in tracing

and measuring environmental externalities from global production and consumption systems adds

complexity to the investigation and implementation of SSCM practices, rendering them often

symbolic (Adhikary et al., 2020; O’Rourke, 2014). Addressing these limitations within the context

of MSCs, our study provides several implications for practitioners.

First, we emphasize the importance of using a comprehensive measure to gauge environmental

performance across supply chains. Operational features may vary depending on inherent and situ-

ational factors of supply chain entities, making it essential to utilize diverse environmental metrics

such as GHG emissions, waste, and environmental costs resulting from operational outcomes to

assess corporate environmental efficiency. A single measure of environmental performance can

obscure these diverse aspects, whereas environmental efficiency scores based on multiple input

and output environmental metrics provide a more accurate reflection of corporate environmental

performance across various dimensions.

Next, we note that separate tier-level models do not consider the linking activities occurring

at different stages of supply chains. Therefore, employing NDEA models in sustainable supply

chain research is effective, as they account for both internal operations and the interconnected

actions of companies within the supply chains. Unlike node-level performance evaluation, network-

level performance evaluation that integrates multifaceted characteristics provides a comprehensive

understanding of how firms at different stages of supply chains interact with members at other

33

tiers. Therefore, using measures that capture multi-dimensional activities and consider cross-tier

interactions will help managers obtain meaningful insights.

Among firms that share their environmental performance data with external stakeholders, our

empirical evidence indicates a positive cross-tier environmental impact. However, the impact may

vary depending on firms’ positions within the supply chain and their business relationships. Specif-

ically, while focal firms can influence the environmental efficiency of their immediate suppliers,

they may not directly affect the sustainable practices of suppliers further upstream. In light of these

findings, even in the absence of direct contractual relationships with upstream suppliers, managers

of focal firms should closely monitor the upstream suppliers’ sustainable practices and implement

mechanisms to incentivize and enhance their environmental efficiency.

Finally, while we find a positive association in cross-tier environmental efficiency, this rela-

tionship varies based on the level of a supplier’s financial dependence in dyadic relationships. Our

empirical evidence indicates that a supplier’s revenue dependence enhances the positive association

between the environmental efficiencies of suppliers and their downstream partners. This indicates

that managers should intensify their efforts to collaborate with suppliers who are less financially

dependent on them to establish and pursue joint environmental goals, as these suppliers may be

less inclined to align their practices with the firm’s environmental standards.

2.6.3 Future Research and Conclusion

Our study has several limitations, giving rise to opportunities for future research. First, to

analyze the impact of cross-tier interactions on tier-level environmental performance, we focus on

open triadic relationships, exploring a specific form of triads. However, the governance mechanisms

in MSCs may be more complex and diverse (Tachizawa and Wong, 2014; Mena et al., 2013; Choi

and Wu, 2009). Therefore, considering other forms of triadic relationships could potentially offer

a more nuanced understanding of the dynamics of environmental performance within MSCs.

Second, to consider the linking activities and revenue dependence between firms at different

tiers of supply chains, we utilize revenue percentage data and relationship longevity of BSRs. In

Section 2.3.2, we detail how the unique features of FactSet Revere supply chain relationships data,

34

including the start and end dates of BSRs, have been handled in previous studies. Notably, these

studies coarsen the start and end dates of BSRs.

In contrast, we use the exact proportions of

relationship durations within a given year. Both approaches may introduce some degree of noise.

Using supply chain relationship data over more granular timespans, such as monthly or quarterly

data, would enhance the robustness of our empirical analyses.

Next, the sample selected for our empirical analysis is exclusively drawn from six distinct

manufacturing industries. While both second-tier and first-tier suppliers operate across a diverse

range of industries, the focal firms in this study have been chosen solely from the manufacturing

sector. As a result, the findings of this research may not be fully generalizable to other sectors.

In conclusion, prior research in SSCM has predominantly focused on managing the sustainability

performance of direct suppliers. However, attention to levels beyond first-tier suppliers has been

limited, often neglecting the dynamics and interactions involving sub-suppliers (Grimm et al.,

2014). Despite certain limitations highlighted earlier, this study advances SSCM research by

examining multiple triadic relationships and employing a unique panel data framework. Our work

sheds light on the dynamics of cross-tier environmental management and highlights the significant

role of supplier revenue dependence in influencing these dynamics.

35

REFERENCES

Adhikary, A., Sharma, A., Diatha, K. S., and Jayaram, J. (2020). Impact of buyer-supplier network
complexity on firms’ greenhouse gas (GHG) emissions: An empirical investigation. International
Journal of Production Economics, 230:107864.

Agca, S., Babich, V., Birge, J. R., and Wu, J. (2022). Credit shock propagation along supply chains:

Evidence from the CDS market. Management Science, 68(9):6506–6538.

Ağca, Ş., Birge, J. R., Wang, Z., and Wu, J. (2023). The impact of COVID-19 on supply chain

credit risk. Production and Operations Management, 32(12):4088–4113.

Al-Tuwaijri, S. A., Christensen, T. E., and Hughes Ii, K. (2004). The relations among environmental
disclosure, environmental performance, and economic performance: A simultaneous equations
approach. Accounting, Organizations and Society, 29(5-6):447–471.

An, Q., Yan, H., Wu, J., and Liang, L. (2016). Internal resource waste and centralization degree in

two-stage systems: An efficiency analysis. Omega, 61:89–99.

Awaysheh, A. and Klassen, R. D. (2010). The impact of supply chain structure on the use of supplier
socially responsible practices. International Journal of Operations & Production Management,
30(12):1246–1268.

Barker, J. M., Hofer, C., Hoberg, K., and Eroglu, C. (2022). Supplier inventory leanness and

financial performance. Journal of Operations Management, 68(4):385–407.

Bates, M. D., Papke, L. E., and Wooldridge, J. M. (2024). Non linear correlated random effects

models with endogeneity and unbalanced panels. Econometric Reviews, pages 1–20.

Bellamy, M. A., Dhanorkar, S., and Subramanian, R. (2020). Administrative environmental in-
novations, supply network structure, and environmental disclosure. Journal of Operations
Management, 66(7-8):895–932.

Bové, A.-T. and Swartz, S. (2016). Starting at the source: Sustainability in supply chains. McKinsey

on Sustainability and Resource Productivity, 4:36–43.

Bozarth, C. C., Warsing, D. P., Flynn, B. B., and Flynn, E. J. (2009). The impact of supply
chain complexity on manufacturing plant performance. Journal of Operations Management,
27(1):78–93.

Brandenburg, M., Govindan, K., Sarkis, J., and Seuring, S. (2014). Quantitative models for
sustainable supply chain management: Developments and directions. European Journal of
Operational Research, 233(2):299–312.

CDP (2019). Cascading Commitments: Driving Ambitious Action through Supply Chain Engage-

36

ment. https://cdn.cdp.net/cdp-production/cms/reports/documents/000/004/072/original/CDP_
Supply_Chain_Report_2019.pdf?1550490556.

Chae, S., Yan, T., and Yang, Y. (2020). Supplier innovation value from a buyer–supplier structural
equivalence view: Evidence from the pace awards in the automotive industry. Journal of
Operations Management, 66(7-8):820–838.

Chen, C. and Yan, H. (2011). Network DEA model for supply chain performance evaluation.

European Journal of Operational Research, 213(1):147–155.

Chen, J., Hu, Q., and Song, J.-S. (2017). Supply chain models with mutual commitments and
implications for social responsibility. Production and Operations Management, 26(7):1268–
1283.

Chen, Y., Cheng, S., and Zhu, Z. (2021). Exploring the operational and environmental perfor-
mance of chinese airlines: A two-stage undesirable SBM-NDEA approach. Journal of Cleaner
Production, 289:125711.

Choi, T. and Linton, T. (2011). Don’t let your supply chain control your business. Harvard Business

Review, 89(12):112–117.

Choi, T. Y. and Wu, Z. (2009). Triads in supply networks: Theorizing buyer–supplier–supplier

relationships. Journal of Supply Chain Management, 45(1):8–25.

Cole, D., Narayanan, S., Connors, E., Tewari, M., and Onda, K. (2023). Water stress: Opportunities

for supply chain research. Production and Operations Management, Special Issue.

Crosignani, M., Macchiavelli, M., and Silva, A. F. (2023). Pirates without borders: The propagation
of cyberattacks through firms’ supply chains. Journal of Financial Economics, 147(2):432–448.

Culot, G., Podrecca, M., Nassimbeni, G., Orzes, G., and Sartor, M. (2023). Using supply chain
databases in academic research: A methodological critique. Journal of Supply Chain Manage-
ment, 59(1):3–25.

De Stefano, M. C. and Montes-Sancho, M. J. (2024). Complex supply chain structures and multi-
scope GHG emissions: The moderation effect of reducing equivocality. International Journal
of Operations & Production Management, 44(5):952–986.

Diebel, W., Gualandris, J., and Klassen, R. D. (2024). How do suppliers respond to institutional
complexity? Examining voluntary public environmental disclosure in a global manufacturing
supply network. Journal of Operations Management, 70(2):285–315.

Dong, Y., Skowronski, K., Song, S., Venkataraman, S., and Zou, F. (2020). Supply base innovation

and firm financial performance. Journal of Operations Management, 66(7-8):768–796.

37

Dreyfus, D., Nair, A., and Talluri, S. (2020). The impact of chain organization size on efficiency
and quality of affiliated facilities—implications for multi-unit organizational forms. Production
and Operations Management, 29(7):1605–1623.

Elking, I., Paraskevas, J.-P., Grimm, C., Corsi, T., and Steven, A. (2017). Financial dependence, lean
inventory strategy, and firm performance. Journal of Supply Chain Management, 53(2):22–38.

Ernst & Young (2022). How sustainable supply chains are driving business transformation. https:
//www.ey.com/en_us/insights/supply-chain/supply-chain-sustainability-2022#:~:text=Survey%
20responses%20from%20525%20senior,primarily%20on%20savings%20and%20efficiency.
Accessed: 2024-07-22.

Fu, W. and Jacobs, B. W. (2022). Does increased water efficiency improve financial performance?
The important role of operational efficiency. International Journal of Operations & Production
Management, 42(3):304–330.

Garcia-Torres, S., Albareda, L., Rey-Garcia, M., and Seuring, S. (2019). Traceability for
sustainability–literature review and conceptual framework. Supply Chain Management: An
International Journal, 24(1):85–106.

Gharfalkar, M., Court, R., Campbell, C., Ali, Z., and Hillier, G. (2015). Analysis of waste hierarchy

in the European waste directive 2008/98/ec. Waste Management, 39:305–313.

Gimenez, C. and Sierra, V. (2013). Sustainable supply chains: Governance mechanisms to greening

suppliers. Journal of Business Ethics, 116:189–203.

Gofman, M., Segal, G., and Wu, Y. (2020). Production networks and stock returns: The role of

vertical creative destruction. The Review of Financial Studies, 33(12):5856–5905.

Gofman, M. and Wu, Y. (2022). Trade credit and profitability in production networks. Journal of

Financial Economics, 143(1):593–618.

Graafland, J. (2019). Economic freedom and corporate environmental responsibility: The role of
small government and freedom from government regulation. Journal of Cleaner Production,
218:250–258.

Grimm, J. H., Hofstetter, J. S., and Sarkis, J. (2014). Critical factors for sub-supplier management:
A sustainable food supply chains perspective. International Journal of Production Economics,
152:159–173.

Gualandris, J., Longoni, A., Luzzini, D., and Pagell, M. (2021). The association between sup-
ply chain structure and transparency: A large-scale empirical study. Journal of Operations
Management, 67(7):803–827.

Gutiérrez, R., Márquez, P., and Reficco, E. (2016). Configuration and development of alliance

38

portfolios: A comparison of same-sector and cross-sector partnerships. Journal of Business
Ethics, 135:55–69.

Hajmohammad, S., Klassen, R. D., and Vachon, S. (2024). Managing supplier sustainability risk:
An experimental study. Supply Chain Management: An International Journal, 29(1):50–67.

Hartmann, J. and Moeller, S. (2014). Chain liability in multitier supply chains? Responsibility
attributions for unsustainable supplier behavior. Journal of Operations Management, 32(5):281–
294.

Hartmann, J. and Uhlenbruck, K. (2015). National institutional antecedents to corporate environ-

mental performance. Journal of World Business, 50(4):729–741.

Hoejmose, S. U., Grosvold, J., and Millington, A. (2013). Socially responsible supply chains:
Power asymmetries and joint dependence. Supply Chain Management: An International Journal,
18(3):277–291.

Hyun, J. and Kim, R. (2018). Propagation of housing market disruptions during the great recession:

Supply chain network channel. Columbia Business School Research Paper, pages 18–63.

Jacobs, B. W., Kraude, R., and Narayanan, S. (2016). Operational productivity, corporate social
performance, financial performance, and risk in manufacturing firms. Production and Operations
Management, 25(12):2065–2085.

Jamalnia, A., Gong, Y., and Govindan, K. (2023). Sub-supplier’s sustainability management in
multi-tier supply chains: A systematic literature review on the contingency variables, and a
conceptual framework. International Journal of Production Economics, 255:108671.

Jira, C. and Toffel, M. W. (2013). Engaging supply chains in climate change. Manufacturing &

Service Operations Management, 15(4):559–577.

Joshi, R. and Wooldridge, J. M. (2019). Correlated random effects models with endogenous explana-
tory variables and unbalanced panels. Annals of Economics and Statistics, GENES(134):243–
268.

Kalaitzi, D., Matopoulos, A., Bourlakis, M., and Tate, W. (2018). Supply chain strategies in an era
of natural resource scarcity. International Journal of Operations & Production Management,
38(3):784–809.

Kao, T.-W. D., Simpson, N., Shao, B. B., and Lin, W. T. (2017). Relating supply network structure
to productive efficiency: A multi-stage empirical investigation. European Journal of Operational
Research, 259(2):469–485.

Keskin, B. (2021). An efficiency analysis on social prosperity: OPEC case under network DEA

slack-based measure approach. Energy, 231:120832.

39

Kim, Y. H. and Henderson, D. (2015). Financial benefits and risks of dependency in triadic supply

chain relationships. Journal of Operations Management, 36:115–129.

Kim, Y. H. and Wemmerlöv, U. (2015). Does a supplier’s operational competence translate into
financial performance? An empirical analysis of supplier–customer relationships. Decision
Sciences, 46(1):101–134.

Kleindorfer, P. R., Singhal, K., and Van Wassenhove, L. N. (2005). Sustainable operations man-

agement. Production and Operations Management, 14(4):482–492.

Martínez-Campillo, A., Wijesiri, M., and Wanke, P. (2020). Evaluating the double bottom-line of
social banking in an emerging country: How efficient are public banks in supporting priority and
non-priority sectors in india? Journal of Business Ethics, 162:399–420.

McFarland, R. G., Bloodgood, J. M., and Payan, J. M. (2008). Supply chain contagion. Journal of

Marketing, 72(2):63–79.

Mena, C., Humphries, A., and Choi, T. Y. (2013). Toward a theory of multi-tier supply chain

management. Journal of Supply Chain Management, 49(2):58–77.

Miller, T., Kim, A. B., Roberts, J. M., and Tyrrell, P. (2022). 2022 Index of Economic Freedom.

Technical report, The Heritage Foundation.

Modi, S. B. and Cantor, D. E. (2021). How coopetition influences environmental performance: Role
of financial slack, leverage, and leanness. Production and Operations Management, 30(7):2046–
2068.

Moreno, P. and Lozano, S. (2014). A network DEA assessment of team efficiency in the NBA.

Annals of Operations Research, 214:99–124.

Mundlak, Y. (1978). On the pooling of time series and cross section data. Econometrica: Journal

of the Econometric Society, pages 69–85.

Osadchiy, N., Schmidt, W., and Wu, J. (2021). The bullwhip effect in supply networks. Management

Science, 67(10):6153–6173.

O’Rourke, D. (2014). The science of sustainable supply chains. Science, 344(6188):1124–1127.

Pankratz, N. M. and Schiller, C. M. (2024). Climate change and adaptation in global supply-chain

networks. The Review of Financial Studies, 37(6):1729–1777.

Papke, L. E. and Wooldridge, J. M. (2008). Panel data methods for fractional response variables

with an application to test pass rates. Journal of Econometrics, 145(1-2):121–133.

Pfeffer, J. and Nowak, P. (1976). Joint ventures and interorganizational interdependence. Adminis-

40

trative Science Quarterly, 21(3):398–418.

Pfeffer, J. and Salancik, G. R. (1978). The external control of organizations: A resource dependence

perspective. Technical report, Harper & Row, New York.

Price, J. L. and Joseph, J. B. (2000). Demand management–a basis for waste policy: A critical review
of the applicability of the waste hierarchy in terms of achieving sustainable waste management.
Sustainable Development, 8(2):96–105.

Ried, L., Eckerd, S., Kaufmann, L., and Carter, C. (2021). Spillover effects of information leakages

in buyer–supplier–supplier triads. Journal of Operations Management, 67(3):280–306.

Sancha, C., Wong, C. W., and Gimenez, C. (2019). Do dependent suppliers benefit from buying
firms’ sustainability practices? Journal of Purchasing and Supply Management, 25(4):100542.

Saunders, L. W., Paul Brooks, J., Merrick, J. R., and Autry, C. W. (2020). Addressing eco-
nomic/environmental sustainability trade-offs in procurement episodes with industrial suppliers.
Production and Operations Management, 29(5):1256–1269.

Schiller, C. (2018). Global supply-chain networks and corporate social responsibility.

In 13th

Annual Mid-Atlantic Research Conference in Finance (MARC) Paper.

Serpa, J. C. and Krishnan, H. (2018). The impact of supply chains on firm-level productivity.

Management Science, 64(2):511–532.

Sodhi, M. S. and Tang, C. S. (2019). Research opportunities in supply chain transparency. Pro-

duction and Operations Management, 28(12):2946–2959.

Song, S., Dong, Y., Kull, T., Carter, C., and Xu, K. (2023). Supply chain leakage of greenhouse gas
emissions and supplier innovation. Production and Operations Management, 32(3):882–903.

Song, S., Lian, J., Skowronski, K., and Yan, T. (2024). Customer base environmental disclosure
and supplier greenhouse gas emissions: A signaling theory perspective. Journal of Operations
Management, 70(3):355–380.

Surroca, J., Prior, D., and Tribo Gine, J. A. (2016). Using panel data DEA to measure CEOs’
focus of attention: An application to the study of cognitive group membership and performance.
Strategic Management Journal, 37(2):370–388.

Tachizawa, E. M. and Wong, C. Y. (2014). Towards a theory of multi-tier sustainable supply
chains: A systematic literature review. Supply Chain Management: An International Journal,
19(5/6):643–663.

Tang, C. S. and Zhou, S. (2012). Research advances in environmentally and socially sustainable

operations. European Journal of Operational Research, 223(3):585–594.

41

Tate, W. L., Ellram, L. M., and Bals, L. (2022). Handbook of Theories for Purchasing, Supply
Chain and Management Research, pages 153–167. Edward Elgar Publishing, Northampton,
MA, 1 edition.

Tate, W. L., Ellram, L. M., and Gölgeci, I. (2013). Diffusion of environmental business practices:

A network approach. Journal of Purchasing and Supply Management, 19(4):264–275.

Tavana, M., Mirzagoltabar, H., Mirhedayatian, S. M., Saen, R. F., and Azadi, M. (2013). A new
network epsilon-based DEA model for supply chain performance evaluation. Computers &
Industrial Engineering, 66(2):501–513.

Tone, K. (2017). Advances in DEA theory and applications. Wiley Online Library.

Tone, K. and Tsutsui, M. (2009). Network DEA: A slacks-based measure approach. European

Journal of Operational Research, 197(1):243–252.

Touboulic, A., Chicksand, D., and Walker, H. (2014). Managing imbalanced supply chain relation-

ships for sustainability: A power perspective. Decision Sciences, 45(4):577–619.

Vedel, M., Holma, A.-M., and Havila, V. (2016). Conceptualizing inter-organizational triads.

Industrial Marketing Management, 57:139–147.

Villena, V. H. and Gioia, D. A. (2018). On the riskiness of lower-tier suppliers: Managing

sustainability in supply networks. Journal of Operations Management, 64:65–87.

Villena, V. H. and Gioia, D. A. (2020). A more sustainable supply chain. Harvard Business Review,

98(2):84–93.

Wang, D. D. (2019). Assessing road transport sustainability by combining environmental impacts
and safety concerns. Transportation Research Part D: Transport and Environment, 77:212–223.

Wang, Y., Li, J., Wu, D., and Anupindi, R. (2021). When ignorance is not bliss: An empirical
analysis of subtier supply network structure on firm risk. Management Science, 67(4):2029–2048.

Wilhelm, M., Blome, C., Wieck, E., and Xiao, C. Y. (2016a).

Implementing sustainability in
multi-tier supply chains: Strategies and contingencies in managing sub-suppliers. International
Journal of Production Economics, 182:196–212.

Wilhelm, M. and Villena, V. H. (2021). Cascading sustainability in multi-tier supply chains: When
do Chinese suppliers adopt sustainable procurement? Production and Operations Management,
30(11):4198–4218.

Wilhelm, M. M., Blome, C., Bhakoo, V., and Paulraj, A. (2016b). Sustainability in multi-tier supply
chains: Understanding the double agency role of the first-tier supplier. Journal of Operations
Management, 41:42–60.

42

Wooldridge, J. M. (2019). Correlated random effects models with unbalanced panels. Journal of

Econometrics, 211(1):137–150.

Wu, J., Yin, P., Sun, J., Chu, J., and Liang, L. (2016). Evaluating the environmental efficiency of a
two-stage system with undesired outputs by a dea approach: An interest preference perspective.
European Journal of Operational Research, 254(3):1047–1062.

Wu, J., Zhang, Z., and Zhou, S. X. (2022). Credit rating prediction through supply chains: A

machine learning approach. Production and Operations Management, 31(4):1613–1629.

Wu, Z., Choi, T. Y., and Rungtusanatham, M. J. (2010). Supplier–supplier relationships in buyer–
supplier–supplier triads: Implications for supplier performance. Journal of Operations Manage-
ment, 28(2):115–123.

Zhou, H., Yang, Y., Chen, Y., and Zhu, J. (2018). Data envelopment analysis application in
sustainability: The origins, development and future directions. European Journal of Operational
Research, 264(1):1–16.

43

APPENDIX

MODELING RELATIVE ENVIRONMENTAL EFFICIENCY IN MSCS USING NDEA

NDEA is a highly effective method for evaluating the efficiencies of various tiers in supply

chains, as it accounts for divisional efficiency at each stage as well as the overall efficiency within

unified supply networks (Tone and Tsutsui, 2009). To comprehensively evaluate bilateral efficiency,

encompassing both individual divisions and the overall system, this study employs a network slack-

based measure (NSBM) approach. A DEA model based on the NSBM approach is suitable when

input and output factors do not change proportionally. Due to the non-radial nature of the NSBM

approach, a variable return-to-scale (VRS) method is used in this study to ensure that at least one

fully efficient DMU is available to serve as a benchmark for other inefficient units (Tone and Tsutsui,

2009). Further, we employ a non-oriented DEA model because firms attempt to achieve a dual

goal that simultaneously aims to decrease the input and increase the output to maximize efficiency

(Keskin, 2021; Martínez-Campillo et al., 2020; Gharfalkar et al., 2015; Moreno and Lozano, 2014;

Price and Joseph, 2000).

In summary, we employ an NDEA model based on the non-oriented

NSBM approach under the VRS assumption.

Building on previous studies that include undesirable outcomes in NDEA to compute efficiencies

(Chen et al., 2021; Martínez-Campillo et al., 2020; Tone and Tsutsui, 2009), we develop an NDEA

model to evaluate tier-level environmental efficiency scores across supply chains, along with the

overall system-level environmental performance. To model the system-level efficiency analysis, we

first describe the data matrices for each input, output, and intermediate linking factor.

In our NDEA model, let 𝑁 be the total number of DMUs and 𝑆 be the number of stages for each

DMU. For the stage 𝑠 ∈ {1, 2, ..., 𝑆} of DMU 𝑛 ∈ {1, 2, ..., 𝑁 }, let 𝑥𝑠

𝑛 denote the vector of inputs

and let 𝑚𝑠 be the number of inputs. Then, we can write the input vector of stage 𝑠 of DMU 𝑛 as

follows:

(cid:16)

𝑥𝑠
𝑛 :=

𝑥𝑠
𝑛,1

, 𝑥𝑠

𝑛,2

, ..., 𝑥𝑠

𝑛,𝑚𝑠

(cid:17)

.

(A.1)

Here, for any 𝑛 ∈ {1, 2, ..., 𝑁 }, 𝑠 ∈ {1, 2, ..., 𝑆}, and 𝑚 ∈ {1, 2, ..., 𝑚𝑠}, the input 𝑥𝑠

𝑛,𝑚 ∈ R+.

44

Similarly, let 𝑦𝑠

𝑛 be the vector and 𝑟𝑠 be the number of desirable outputs of stage 𝑠 of DMU 𝑛,

respectively. Then, we write the vector of desirable outputs of stage 𝑠 of DMU 𝑛 as follows:

(cid:16)

𝑦𝑠
𝑛 :=

𝑦𝑠
𝑛,1

, 𝑦𝑠

𝑛,2

, ..., 𝑦𝑠

𝑛,𝑟𝑠

(cid:17)

.

(A.2)

Next, let ˜𝑦𝑠

𝑛 be the vector and 𝑡𝑠 be the number of undesirable outputs of stage 𝑠 of DMU 𝑛,

respectively. Then, the vector of undesirable outputs of stage 𝑠 of DMU 𝑛 is:

(cid:16)

˜𝑦𝑠
𝑛 :=

˜𝑦𝑠
𝑛,1

, ˜𝑦𝑠
𝑛,2

, ..., ˜𝑦𝑠

𝑛,𝑡𝑠

(cid:17)

.

(A.3)

Finally, for any 𝑠1 ∈ {1, 2, ..., 𝑆} and 𝑠2 = (𝑠1 + 1), let 𝑧(𝑠1,𝑠2)

be the vector and 𝑝𝑠1 be the
number of intermediate variables linking stage 𝑠1 to stage 𝑠2 of DMU 𝑛, respectively. We then

𝑛

define the vector of intermediate variables linking stage 𝑠1 to stage 𝑠2 of DMU 𝑛 as follows:

𝑧(𝑠1,𝑠2)
𝑛

:=

(cid:16)

𝑧(𝑠1,𝑠2)
𝑛,1

, 𝑧(𝑠1,𝑠2)
𝑛,2

, ..., 𝑧(𝑠1,𝑠2)

𝑛,𝑝𝑠
1

(cid:17)

.

(A.4)

Here, for any 𝑛 ∈ {1, 2, ..., 𝑁 }, 𝑠 ∈ {1, 2, ..., 𝑆}, 𝑟 ∈ {1, 2, ..., 𝑟𝑠}, 𝑡 ∈ {1, 2, ..., 𝑡𝑠}, and

𝑛,𝑡, and 𝑧(𝑠,𝑠+1)
𝑝 ∈ {1, 2, ..., 𝑝𝑠1 }, the variables 𝑦𝑠
R+. Using these input, output, and intermediate linking factors, similar to Tone (2017); Tone and

belong to the set of positive real numbers

𝑛,𝑟, ˜𝑦𝑠

𝑛,𝑝

Tsutsui (2009), we build our NDEA optimization model as follows:

Observe that the model in (A.5) is an NSBM model with both desirable and undesirable outputs.

The decision variables ˆ𝛿𝑠−, 𝛿𝑠, and ˜𝛿𝑠 represent the amount of slack in the inputs, desirable outputs,

and undesirable outputs, respectively. The decision variables 𝜆𝑠 refer to weights assigned to the

DMUs corresponding to stage 𝑠 (Tone, 2017). The parameter 𝑊 𝑠 represents the weight assigned to

the efficiency of each stage. Consistent with previous studies (Chen et al., 2021; Martínez-Campillo

et al., 2020; Tone and Tsutsui, 2009), we assign equal weight to this parameter in our model. We

represent the optimal overall efficiency as 𝜌∗
𝑜.

45

(cid:18) 𝑚𝑠
(cid:205)
𝑚=1

(cid:19)(cid:21)

ˆ𝛿𝑠−
𝑚
𝑥𝑠
𝑚,𝑜

(cid:20)

𝑊 𝑠

𝑆
(cid:205)
𝑠=1
(cid:20)

𝑊 𝑠

1 +

1 − 1
𝑚𝑠
(cid:17) (cid:18) 𝑟𝑠
(cid:205)
𝑟=1

(cid:16) 1
𝑟𝑠+𝑡𝑠

𝛿𝑠
𝑟
𝑦𝑠
𝑟 ,𝑜

+

(cid:19)(cid:21)

𝑡𝑠
(cid:205)
𝑡=1

˜𝛿𝑠
𝑡
˜𝑦𝑠
𝑡 ,𝑜

min

𝜌∗
𝑜
𝜆𝑠, ˆ𝛿𝑠−,𝛿𝑠, ˜𝛿𝑠

=

𝑆
(cid:205)
𝑠=1

subject to

𝑥𝑠
𝑚,𝑜 =

𝑦𝑠
𝑟,𝑜 =

˜𝑦𝑠
𝑡,𝑜 =

𝑁
∑︁

𝑛=1
𝑁
∑︁

𝑛,𝑚𝜆𝑠
𝑥𝑠

𝑛 + ˆ𝛿𝑠−
𝑚

∀1 ≤ 𝑚 ≤ 𝑚𝑠, 1 ≤ 𝑠 ≤ 𝑆,

𝑛,𝑟𝜆𝑠
𝑦𝑠

𝑛 − 𝛿𝑠
𝑟

∀1 ≤ 𝑟 ≤ 𝑟𝑠, 1 ≤ 𝑠 ≤ 𝑆,

𝑛=1
𝑁
∑︁

𝑛=1

𝑛,𝑡𝜆𝑠
˜𝑦𝑠

𝑛 + ˜𝛿𝑠
𝑡

∀1 ≤ 𝑡 ≤ 𝑡𝑠, 1 ≤ 𝑠 ≤ 𝑆,

(A.5)

𝑁
∑︁

𝑛=1
𝑁
∑︁

𝑛=1
𝑆
∑︁

𝑧(𝑠,𝑠+1)
𝑛,𝑝

𝜆𝑠
𝑛 =

𝑁
∑︁

𝑛=1

𝑧(𝑠,𝑠+1)
𝑛,𝑝

𝜆𝑠+1
𝑛

∀1 ≤ 𝑝 ≤ 𝑝𝑠, 1 ≤ 𝑠 ≤ 𝑆,

𝑛 = 1 ∀1 ≤ 𝑠 ≤ 𝑆, 𝜆𝑠
𝜆𝑠

𝑛 ≥ 0 ∀1 ≤ 𝑠 ≤ 𝑆, 1 ≤ 𝑛 ≤ 𝑁,

𝑊 𝑠 = 1,

𝑠=1
𝑚 , 𝛿𝑠
ˆ𝛿𝑠−

𝑟, ˜𝛿𝑠

𝑡 , 𝑊 𝑠 ≥ 0 ∀1 ≤ 𝑠 ≤ 𝑆, 1 ≤ 𝑚 ≤ 𝑚𝑠, 1 ≤ 𝑟 ≤ 𝑟𝑠, 1 ≤ 𝑡 ≤ 𝑡𝑠.

46

CHAPTER 3

ENVIRONMENTAL DISCLOSURE IN SUPPLY NETWORKS

3.1

Introduction

In recent years, consumers have grown increasingly interested in how the products they use

are sourced and produced. According to recent research, consumers may be inclined to spend

an additional 2% to 10% on products from companies that offer enhanced transparency in their

supply chains (Bateman and Bonanni, 2019; Kraft et al., 2018). Faced with increasing pressures

from diverse stakeholders, firms tend to put in extra effort to shift the boundaries of supply chain

information disclosure (Marshall et al., 2016). For example, Patagonia launched a Supply Chain

Environmental Responsibility Program aimed at evaluating, reducing, and ultimately eliminating

the environmental impact associated with the production of its products and materials. Patagonia

routinely assesses the manufacturing facilities of current and prospective suppliers to form a shared

supply chain, and an increasing number of companies participate in these initiatives to assess and

disclose their environmental impacts.

Supplier sustainability actions can also be tracked and disclosed in compliance with government

regulations. The Corporate Sustainability Reporting Directive (CSRD) took effect on January 5,

2023, updating and reinforcing the regulations regarding the reporting of social and environmental

information by companies. Specifically, this European Union (EU) regulation mandates that all

large and publicly listed small and midsize enterprises (SMEs) must report on perceived risks and

opportunities related to social and environmental issues, as well as the effects of their operations on

people and the environment. With the new rules, both large companies and SMEs, often upstream

suppliers in supply chains, are required to regularly report on social and environmental risks. This

enables investors and stakeholders to access data that evaluates the impact of companies on people

and the environment, as well as to analyze the financial risks and opportunities related to climate

change and other sustainability issues.

In the context of sustainable supply chain management (SSCM), previous literature has high-

lighted two key practices for information sharing within supply chains: supply chain transparency

47

and supply chain visibility. Supply chain visibility is often defined as the degree to which a

company can have access to the information within its supply chain (Oracle Corporation, 2023;

Sodhi and Tang, 2019; Kraft et al., 2018; Basole and Bellamy, 2014). On the other hand, supply

chain transparency means a company disclosing information to the external stakeholders, including

consumers and investors, about upstream and downstream operations (Mollenkopf et al., 2022;

Gualandris et al., 2021; Sodhi and Tang, 2019; Pagell and Wu, 2009).

In other words, supply

chain transparency requires companies to comprehend activities happening in their supply chain

and to communicate this information internally within the organization as well as externally to

stakeholders (Gualandris et al., 2021; Bateman and Bonanni, 2019). Although both concepts aim

to convey information for supply chain management (SCM), firms disclose information to varying

degrees based on their varying perceptions of the value of supply chain visibility and transparency.

Disclosure is often defined as a company’s decision for the purpose of information sharing and

communication with diverse stakeholders (Kraft et al., 2018; Gualandris et al., 2015). In general,

firms are determined to disclose their information to the public to gain credibility and legitimacy,

which is defined as “a generalized perception or assumption that the actions of an entity are

desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs,

and definitions (Delgado-Márquez et al., 2017; Haniffa and Cooke, 2005; Mitchell et al., 1997).”

Therefore, both stakeholders (such as regulators, investors, and governments) and non-stakeholders

(including the general public who may not have a direct interest) influence and are interested in the

information that is disclosed to the public (Jira and Toffel, 2013).

Firms disclose their environmental efforts not only to comply with regulatory requirements

but also to protect their reputation among increasingly environmentally conscious stakeholders.

These disclosures serve as signals of internal sustainability commitments, shaping how investors,

regulators, and consumers perceive a firm’s social responsibility.

Environmental disclosure has garnered substantial attention in the context of SCM and buyer–supplier

relationships (BSRs) (Bellamy et al., 2020; Gualandris et al., 2015; Jira and Toffel, 2013). Broadly

defined, it reflects the extent to which a firm publicly reports its environmental impacts, policies,

48

and practices (Diebel et al., 2024; Song et al., 2024; Bellamy et al., 2020). Amid increasing

societal and regulatory scrutiny, stakeholders are demanding greater transparency regarding firms’

environmental footprints, particularly as these affect communities embedded within global supply

chains (Gualandris et al., 2021; Bellamy et al., 2020; Tate et al., 2010).

In response, firms’ engagement in environmental, social, and governance (ESG) initiatives,

particularly environmental disclosure, is shaped not only by internal strategic priorities but also by

their network positions and inter-organizational relationships. Suppliers’ willingness to disclose

environmental information often hinges on the behavior and expectations of their buyers, including

how buyers signal, request, or utilize such information in relational exchanges (Diebel et al., 2024;

Villena and Dhanorkar, 2020; Jira and Toffel, 2013). Beyond these dyadic dynamics, a range of

firm- and context-specific factors, such as environmental performance, administrative innovation,

national institutional context, industry characteristics, and supply chain visibility, also influence

disclosure practices across networked firms (Diebel et al., 2024; Gualandris et al., 2021; Kraft and

Zheng, 2021; Bellamy et al., 2020; Pucheta-Martínez and Gallego-Álvarez, 2020; Jira and Toffel,

2013; Cho et al., 2012; Scholtens and Dam, 2007). Accordingly, supplier environmental disclosure

emerges as a complex outcome shaped by both buyer influence and broader structural and relational

dimensions within supply networks.

Buying firms often play a central role in shaping their suppliers’ business strategies and influenc-

ing key operational decisions. While traditional supply chain research has focused on linear, dyadic

relationships between buyers and suppliers (Cox et al., 2001; Zhu and Sarkis, 2004; Kim et al.,

2011), this perspective overlooks the more complex and interconnected nature of buyer–supplier

networks.

In practice, corporate strategy and environmental decision-making are embedded in

broader systems of interactions involving not just direct partners but also shared ties and indirect

influences. As such, BSRs are frequently better conceptualized as dynamic networks rather than

simple transactional links (Kim et al., 2011).

Applying a social capital perspective provides a more nuanced understanding of how the char-

acteristics of these networked relationships moderate inter-firm dynamics. Social capital, defined

49

as “the actual and potential resources embedded within, available through, and derived from the

network of relationships possessed by [...] a social unit” (Nahapiet and Ghoshal, 1998), facilitates

the flow of information, trust, and norms across supply networks. Within the context of environ-

mental disclosure, social capital embedded in BSRs plays a critical role in shaping how buyers

exert influence and how suppliers respond. Building on prior research, we argue that three dimen-

sions of social capital are especially salient in determining whether buyer influence encourages,

discourages, or has no significant effect on supplier environmental disclosure, particularly in the

manufacturing sector where environmental pressures and disclosure expectations are increasingly

diffused across the network.

This paper first investigates how the environmental disclosure of buyers operating in the man-

ufacturing industry influences the level of environmental disclosure among their suppliers. While

conventional wisdom holds that greater transparency by buyers should encourage suppliers to fol-

low suit, emerging evidence related to the supply chain leakage phenomenon and the pollution

haven hypothesis (PHH) suggests an alternative possibility. Our second research question explores

how the inherent network characteristics of BSRs enhance relational cohesion, thereby influencing

the extent to which suppliers engage in voluntary environmental disclosure. That is, the negative

influence proposed by the supply chain leakage phenomenon and the PHH may be either amplified

or mitigated under specific conditions of structural equivalence, relationship duration, and cultural

similarity, depending on the strength of supply chain visibility with suppliers. Further, we intend

to explore potential disparities between the materials industry and the pharmaceutical and biotech-

nology industry, despite their joint classification under the manufacturing sector in prior research

(Diebel et al., 2024; Gualandris et al., 2021).

3.1.1 Key Findings and Contributions

Our research contributes to the environmental management and SCM literature. This study

emphasizes the role of social capital within BSRs in shaping supplier environmental disclosure,

particularly in light of supply chain leakage concerns. In the manufacturing sector, many suppliers

remain hesitant to disclose their environmental performance – a pattern we interpret through the

50

lens of the PHH and supply chain leakage concepts.

We further explore how firm-level and network-level characteristics influence the link between

buyer and supplier environmental disclosures. Grounded in social capital theory, we examine the

moderating effects of three key dimensions – structural, relational, and cognitive dimensions –

within BSRs in the materials and pharmaceutical industries. Using voluntary disclosure data from

Bloomberg ESG, we assess the relationship between buyer and supplier environmental disclosure.

We find no evidence supporting a negative influence, suggesting that the supply chain leakage

effect may not be broadly applicable in the context of voluntary environmental disclosure. Our

analysis shows that structural social capital, reflected in shared network ties, plays a key moderating

role in this relationship. Buyer–supplier pairs with lower structural overlap experience a negative

effect, whereas pairs with higher overlap show a mitigation of this negative influence. A similar

conditional pattern emerges for cultural similarity: partnerships characterized by greater cultural

dissimilarity demonstrate a negative effect on supplier disclosure, while those with higher cultural

similarity show no significant effect.

In contrast, the length of the BSR does not significantly

moderate the relationship.

This study contributes to the theoretical understanding of how BSRs, shaped by dimensions of

social capital, influence supplier environmental disclosure within manufacturing supply networks.

While previous research has largely focused on buyer performance, we examine how supply chain

leakage and the PHH relate to suppliers’ disclosure behaviors. We also highlight how relational

factors, such as structural equivalence and cultural similarity, positively moderate the influence be-

tween buyer and supplier environmental disclosures. This implies that network dynamics can offset

the potential negative impacts of outsourcing. From a managerial perspective, our findings un-

derscore the importance of fostering cooperative, strategically aligned partnerships with suppliers,

particularly those that share common connections or cultural backgrounds, to enhance environ-

mental transparency. Buyers should also be mindful of how their network positions influence

supplier behavior and prioritize close collaboration, especially when working with culturally di-

verse partners. Policymakers, in turn, should recognize the heterogeneity of firms across industries

51

and supply chains, and tailor sustainability initiatives that consider local contexts and relational

dynamics. Together, these insights offer a more nuanced understanding of how environmental

responsibility can be promoted across interconnected supply networks.

3.1.2 Organization

Our paper is organized as follows: We begin by reviewing the relevant literature on transparency

within SSCM and the institutional factors emphasized in corporate social responsibility (CSR) in

the upcoming section. Following this, Section 3 delves into our hypotheses, which we develop

through a detailed examination of the contextual background and social capital literature. This

section also introduces our conceptual framework for testing these hypotheses. In Section 4, we

outline our research design, detailing the data sources, sample selection process, and the definitions

of our variables. This section further explains our estimation models and discusses the findings.

Section 5 extends our analysis with additional insights through comparisons across buyer industries

and includes robustness checks that employ various empirical specifications. Finally, Section 7

wraps up the paper by discussing the implications of our findings and suggesting avenues for future

research.

3.2 Literature Review

Scholars and practitioners are increasingly focusing on sustainability within supply chains, pri-

marily driven by the escalating demand for transparency in supply chain practices. In this section,

we will review existing literature on environmental management and disclosure within supply net-

works. In addition, we plan to explore the characteristics of supply networks within BSRs and their

impact on firm-level environmental performances and actions. Following a comprehensive exam-

ination of sustainability in supply networks, we will shift our focus to the theoretical background

and develop and test a series of hypotheses.

Evaluating sustainability performance often extends beyond the activities of any single tier

or firm. This is because firms are intrinsically part of broader supply networks that include up-

stream suppliers and buying firms, facilitating the exchange of products, services, and information

across diverse geographical locations. In addition, sustainability in supply chains not only mea-

52

sures traditional operational outcomes, such as profit and loss, but also encompasses an expanded

conceptualization of performance that includes social and environmental dimensions (Koberg and

Longoni, 2019; Pagell and Wu, 2009; Elkington and Rowlands, 1999). Environmental management

in supply networks has often been addressed through the adoption of environmental management

systems and practices, such as International Organization for Standardization (ISO) 14000 certi-

fications and eco-management audit scheme, aiming to monitor and control the impact of firm

operations on the natural environment (Hardcopf et al., 2019; González et al., 2008; Montabon

et al., 2007). As such, environmental management falls within the wider scope of sustainability,

which has garnered growing interest in supply chain and operations management (OM) in conjunc-

tion with the release of voluntary and international environmental standards (Hofer et al., 2012;

Linton et al., 2007). Moreover, being environmentally proactive is closely associated with new

business opportunities, efficient resource utilization, and an enhanced corporate image, all of which

can yield significant benefits (Montabon et al., 2007).

Along with the growing interest in environmental management in supply chains, corporate

environmental disclosure has long been the focus of scholarly research. Historically, corporate

environmental reporting has served as a voluntary tool for companies to disclose their environmental

impacts and potential risks, aimed at addressing the needs of their key stakeholders (Fallan, 2016;

Gualandris et al., 2015). Such reporting has become more institutionalized, serving not only to

monitor environmental impacts but also to achieve environmental goals that include enhancing

accountability and transparency (Christensen et al., 2021). Nonetheless, disclosure often ranges

from little information to comprehensive information on sustainability evaluation, verification

processes, and outcomes (Gualandris et al., 2015).

Stakeholders and the public have pressurized firms to disclose the environmental impacts of

their business operations on the communities across their supply chains (Gualandris et al., 2021;

Bellamy et al., 2020; Tate et al., 2010). Firms often showcase their corporate responsibility

performance as a means of enhancing the public image as well as achieving CSR goals. In the

context of corporate environmental disclosure, socio-political theories like legitimacy theory and

53

stakeholder theory have been used to explain how firms respond to institutional pressure by providing

more comprehensive environmental information. Legitimacy theory provides the rationale behind

corporate environmental disclosure in response to social and political pressure. To be specific,

compared to economic legitimacy monitored through the marketplace, social legitimacy is more

likely to be monitored via the policy process by actively participating in the public policy processes,

one of which is disclosure (Patten, 2002). On the other hand, stakeholder theory can also be applied

to the problems of CSR, and it also helps to identify the different situations in which stakeholders

with varying degrees of power, legitimacy, and urgency are presented (Tate et al., 2022; Mitchell

et al., 1997). Drawing on institutional theory, previous research has also considered the institutional

pressures from buyers and industry peers that compel suppliers to disclose significant environmental

impacts and risks in supply networks (Diebel et al., 2024; Villena and Dhanorkar, 2020).

As a contingency factor, the institutional context has been highlighted in prior studies that

emphasize the disparities in CSR reporting between countries. These studies suggest that differences

in national contextual factors could account for the variation in CSR disclosure. The reasons for

social and environmental reporting requirements differ between these countries, influenced by

external forces, key stakeholders, significant investments, and long-term commitments, leading

to discrepancies (Luo et al., 2013). Cultural difference across countries also strongly affects

ethics, CSR, organizational culture, and managerial practices (Pucheta-Martínez and Gallego-

Álvarez, 2020; Scholtens and Dam, 2007). Along these lines, multinational corporations (MNCs)

tend to exhibit greater environmental consciousness when operating in developed countries, where

regulatory standards are typically more stringent. In contrast, operations in less-developed countries

are often subject to more lenient environmental regulations (Hassan, 2023; Eskeland and Harrison,

2003) and are more prone to financial constraints that can hinder the implementation of sustainability

initiatives (Luo et al., 2013). Thus, as part of corporate strategic decision-making, firms often

choose to locate manufacturing facilities in countries with more lenient environmental regulations to

minimize pollution abatement costs (Berry et al., 2021). For instance, strict regulatory frameworks

in developed countries often incentivize the relocation of pollution-intensive industries to less

54

developed regions, a dynamic commonly explained by the PHH (Levinson and Taylor, 2008;

Eskeland and Harrison, 2003).

Along with the national context, industry characteristics are also commonly employed to elu-

cidate the scope and content of social and environmental disclosure (Reverte, 2009; Brammer and

Pavelin, 2008; Cowen et al., 1987). The influence of peers within the same industry on corporate

decision-making has been discussed in previous studies (Zhao and Wang, 2024; Huang et al., 2023;

Lin et al., 2018; Leary and Roberts, 2014). In the realm of CSR, it is recognized that managers

within the same industry are incentivized to emulate the CSR decisions of their peers, highlighting

the significant role industry peers play in enhancing firms’ CSR performance (Zhao and Wang,

2024; Chen et al., 2023; Huang et al., 2023; Campbell, 2007). In addition, the average behavior of

peer groups within the same industry influences corporate disclosure practices, leading to reduced

environmental uncertainty and improved quality of managerial private information (Seo, 2021).

Our research examines the social capital factors embedded within supply networks in the

context of environmental disclosure. In practice, MNCs that monitor and manage their suppliers

are increasingly prioritizing sustainable operations. They pursue this by restructuring their supply

chain architectures, redefining partnerships, and even collaborating with competitors to improve

scale efficiency (Lee, 2010). Since corporate strategy and firms’ network behavior are heavily

shaped by the structure and complexity of supply networks, it is crucial to understand how suppliers,

buyers, and customers are interconnected to fully capture the dynamics of their transactions (Kim

et al., 2011). Moreover, the relationships between suppliers and buyers are plausibly influenced

by the broader characteristics of the supply network, which can either constrain or enhance firm

performance within these interconnected systems (Bellamy et al., 2020; Chae et al., 2020; Kumar

et al., 2020; Sharma et al., 2020; Lu and Shang, 2017).

Over the past few decades, many MNCs have increasingly sought business partners across

diverse geographic regions. Faced with evolving business environments and growing institutional

pressures to operate sustainably, these firms have intensified efforts to strengthen relationships

with their partners in pursuit of environmental objectives. Adopting a social capital perspective,

55

this study examines how the negative influence of buyer environmental disclosure on supplier

environmental disclosure is conditioned by supply chain visibility, which is often shaped by key

dimensions of social capital. By examining how social capital embedded in buyer–supplier dyads

either bridges or exacerbates information gaps within supply networks, this research contributes to

the literature on BSRs and environmental management through a social capital lens.

3.3 Hypotheses Development and Conceptual Framework

In this section, we first formulate a hypothesis regarding corporate environmental disclosure in

BSRs. Building on this foundational hypothesis, we proceed to examine the moderating influence

of social capital dimensions, with a specific focus on the features embedded in BSRs within supply

networks.

3.3.1 Buyer Environmental Disclosure and Its Impact on Supplier Environmental Disclosure

Companies often prioritize different dimensions of social, environmental, and economic re-

sponsibility within their supply chains, depending on institutional factors such as national and

industry contexts (Tate et al., 2010; Van der Laan Smith et al., 2005). As part of their strategic

decision-making, firms may choose to locate manufacturing facilities in countries with more lenient

environmental regulations to reduce pollution abatement costs (Berry et al., 2021). For example,

the stringent regulatory environment in developed countries prompts the relocation of polluting

industries to less developed regions (Levinson and Taylor, 2008; Eskeland and Harrison, 2003).

These outsourced operational activities often enable firms to leverage complementary resources,

thereby enhancing operational efficiency and financial performance, while such benefits may also

extend to improvements in the supply chain’s environmental performance (Song et al., 2023). This

phenomenon is succinctly captured by the PHH. The PHH illustrates a form of institutional or juris-

dictional arbitrage, whereby firms strategically locate their operations in institutional environments

that offer more lenient regulatory conditions (Berry et al., 2021).

Within the context of SCM, supply chain leakage manifests as a firm-level phenomenon, yet it

bears a resemblance to the PHH (Song et al., 2023). Previous literature on environmental manage-

ment within supply chains has explored the concept of supply chain leakage, particularly in terms of

56

outsourcing emissions, commonly known as carbon leakage. Outsourcing emissions to the supply

chain can occur when firms seek to preserve their reputation and social capital without making

significant efforts to respond to pressures from local and national institutions (Song et al., 2024).

This might lead to an escalation in suppliers’ greenhouse gas (GHG) emissions, as certain cus-

tomers might attempt to lower their own emissions at their suppliers’ expense, a practice commonly

known as carbon outsourcing or leakage. Consequently, enhanced environmental disclosure by

customers could inadvertently lead to undesirable environmental performance among their suppli-

ers, suggesting that environmental disclosure might prompt some opportunistic partners to engage

in carbon outsourcing, which potentially creates a negative externality across supply chains (Song

et al., 2024).

To delve deeper into how a firm’s environmental actions, initiatives, efforts, or encouragement

can influence environmental disclosure, and to explore the potential positive relationship between

environmental performance and disclosure levels, we review existing literature on the intercon-

nections between environmental disclosure and both environmental and economic performance.

It has been demonstrated that more extensive disclosure of environmental information correlates

positively with environmental performance, which may also significantly enhance economic per-

formance (Al-Tuwaijri et al., 2004; Cormier and Magnan, 1999). Specifically, firms with higher

performance levels are more likely to disclose pollution-related environmental information com-

pared to their lower-performing counterparts (Clarkson et al., 2008; Al-Tuwaijri et al., 2004). This

suggests that firms use environmental disclosure as a means to communicate with market partic-

ipants, offering transparent, environment-related information as positive news, while also setting

a rational baseline for future reference. Thus, in line with voluntary disclosure theory, firms that

exhibit superior sustainability performance often opt for high-quality sustainability disclosures as a

means to signal their exemplary sustainability achievements (Rezaee and Tuo, 2017; Hummel and

Schlick, 2016).

Suppliers often disclose their environmental efforts not only to comply with regulatory mandates

but also to protect their reputation among increasingly environmentally conscious stakeholders. In

57

many cases, suppliers draw on information shared by their buyers to mirror the buyers’ commit-

ment to transparency, often driven by institutional pressure and stakeholder expectations (Diebel

et al., 2024; Villena and Dhanorkar, 2020; Jira and Toffel, 2013). National culture and business

environments also play a critical role in shaping communication between firms and stakeholders,

as well as in defining the institutional context within which firms, governments, and civil society

interact (Villena and Dhanorkar, 2020; Jira and Toffel, 2013; Van der Laan Smith et al., 2005).

Despite these dynamics, emerging evidence on the supply chain leakage phenomenon and the

PHH points to a countervailing possibility. When a focal firm reduces its carbon emissions and

promotes a favorable environmental image through disclosure, it may do so by shifting pollution-

intensive operations upstream to suppliers. This leakage of carbon can place a disproportionate

environmental burden on suppliers, raising their carbon footprints. Since firms tend to disclose

environmental data when performance is strong, suppliers with higher emissions and wastes may

become reluctant to report such information, resulting in lower levels of disclosure. Building

on prior research into supply chain leakage, which shows that local environmental transparency

or regulatory enforcement can produce unintended consequences, and considering the theoretical

underpinnings of this dynamic, we propose the following hypothesis:

Hypothesis 1 (𝐻1) A buyer’s environmental disclosure negatively influences its supplier’s

environmental disclosure.

3.3.2 The Impact of Social Capital on Environmental Disclosure in Supply Chains

In this section, we mainly explore how the different dimensions of social capital embedded in

BSRs moderate the influence between buyer and supplier environmental disclosure. We begin by

establishing the theoretical foundation of the study, drawing on social capital theory as the primary

lens through which the research hypotheses are developed. Next, we examine how the structural,

relational, and cognitive dimensions of social capital individually shape the relationship between

buyer and supplier environmental disclosure levels. Subsequently, we review previous studies to

formulate a series of hypotheses for empirical evaluation, examining how these three dimensions

influence the relationship between a supplier’s environmental disclosure and that of a buying firm.

58

Behaviors within a dyad ripple through the network, initiating a sequence of actions and re-

sponses that establish norms and expectations across the entire network (Ireland and Webb, 2007).

Thus, cohesion within the network plays a vital role in supporting information exchange and foster-

ing a common identity, leading to opportunities for joint learning and the adoption of sustainable

and social initiatives across supply chains. Through this section, we, therefore, seek to deepen

the understanding of how buying firms can strengthen cohesion with suppliers by leveraging the

social capital embedded in buyer-supplier networks, thereby enhancing supplier engagement in

sustainability practices, including environmental disclosure. Our theoretical framework proposes

that suppliers’ decisions to disclose non-financial information are shaped not only by the environ-

mental disclosure practices of buyers in manufacturing industries but also by the degree of supply

chain visibility operationalized through social capital factors embedded within BSRs. Specifically,

we highlight how the diversity in buyers’ disclosure levels within these networks shapes suppliers’

disclosure strategies.

From a network perspective, social capital underscores how economic behavior is embedded

in interpersonal relationships and structural positions within networks (Lin, 2017; Granovetter,

1985). It emphasizes the value of diverse connections that facilitate trust, cooperation, and the

mobilization of resources through collaboration (Blount and Li, 2021; Kim, 2014; Autry and

Griffis, 2008; Nahapiet and Ghoshal, 1998). Central to this concept are structural and relational

embeddedness, where actors leverage their positions for anticipated returns (Burt, 2004). Positional

advantages, such as occupying structural holes, enhance access to non-redundant information and

resources (Lu and Shang, 2017; Podolny and Baron, 1997; Burt, 1992), while closed networks and

strong ties foster cooperation, norm adherence, and overlapping inter-organizational connections

(Moran, 2005; Granovetter, 1973).

In the context of social networks among individuals or social units, social capital has been

discussed in terms of three primary aspects: structural, relational, and cognitive dimensions (Blount

and Li, 2021; Chae et al., 2020; Nahapiet and Ghoshal, 1998). In previous literature, the structural

dimension is primarily defined as the properties of the social system and overall patterns within

59

network relations, whereas the relational dimension is referred to as the interpersonal or inter-

organizational linkages that entities have developed through interactions. In addition, the cognitive

dimension indicates “those resources providing shared representations, interpretations, and systems

of meaning among parties (Nahapiet and Ghoshal, 1998, p.244).” In Table 3.1, we primarily

focus on studies exploring the dimensions of social capital as the theoretical foundation for their

analyses. Furthermore, Ravindran et al. (2015) extended the three dimensions of social capital to

examine the impact of the four distinct levels of client and vendor firms’ embeddedness of economic

relationships in which firms’ social capital in an inter-organizational network was operationalized by

the four distinct measures of embeddedness: structural embeddedness at the node level, relational

embeddedness at the dyad level, contractual embeddedness at the level of a neighborhood of

contracts, and positional embeddedness at the level of the entire network (Ravindran et al., 2015).

Social capital has often been used alongside both dyadic and network perspectives to evaluate how

its various dimensions impact different aspects of corporate performance within supply networks

(Kumar et al., 2020).

As corporate environmental reporting becomes institutionalized as a tool for promoting account-

ability and transparency, firms increasingly disclose their environmental performance in response

to institutional pressures and stakeholder demand, as a signal of their commitment to sustainability

(Bellamy et al., 2020; Jira and Toffel, 2013). In SCM, transparent buyer disclosure often encourages

suppliers to follow suit, fostering shared sustainability goals. Building on this, the following section

explores how social capital, through its interaction with network structures and its role in shaping

supply chain visibility, moderates the relationship between buyers’ and suppliers’ environmental

disclosures.

3.3.2.1 The Influence of Social Capital’s Structural Dimension

Even though there is no clear consensus concerning how members in a social network pursue

the values derived from social capital, the total values of structural components (e.g., relative

social position, distance between members and existing connections) often represents the structural

dimension of social capital in a social network (Autry and Griffis, 2008).

In other words, the

60

Table 3.1 A Review of Previous Work and Their Use of Three Social Capital Dimensions

Reference

Research Focus

Social Capital Dimensions

Nahapiet and Ghoshal (1998) The creation of intellectual capital

Tsai and Ghoshal (1998)

Patterns of resource exchange and
product innovation

Autry and Griffis (2008)

Supply chain knowledge
development and performance

Villena et al. (2011)

Value creation in BSRs

Kim (2014)

A buying firm’s operational and
financial performance

Claridge (2018)

Dimensions of social capital

Chae et al. (2020)

Supplier innovation value

Blount and Li (2021)

Buyers’ procurement activities
with ethnic minority businesses
and supplier diversity

Wang et al. (2023)

Green innovation output

• Structural dimension
• Relational dimension
• Cognitive dimension

• Structural dimension
• Relational dimension
• Cognitive dimension

• Structural dimension
• Relational dimension

• Structural dimension
• Relational dimension
• Cognitive dimension

• Structural dimension
• Relational dimension

• Structural dimension
• Relational dimension
• Cognitive dimension

• Structural dimension
• Relational dimension
• Cognitive dimension

• Structural dimension
• Relational dimension
• Cognitive dimension

• Structural dimension
• Relational dimension
• Cognitive dimension

structural dimension of social capital mainly focuses on the relative positions within a network and

the continuous formation of relationships among its members. For example, structural density (i.e.,

the extent to which possible connections are activated) and structural holes (i.e., less dense areas in

which ties are missing; see e.g., Burt (1992)) are the two dominant forms of structural embeddedness

61

within a network (Autry and Griffis, 2008).

In particular, dense networks reduce information

asymmetry and opportunism, allowing firms to better understand their partners’ behavior. This

not only fosters increased cooperation but also mitigates opportunistic behavior (Bellamy et al.,

2020; Phelps, 2010). Information flow and knowledge spillovers can also be enhanced by locally

dense interactions connected via a few bridging ties (i.e., small-world networks, Watts and Strogatz

(1998)) (Fleming et al., 2007). Thus, structural components, such as a firm’s network position and

the configuration and distribution of embedded exchange relationships, play a crucial role in the

economic actions of firms (Autry and Griffis, 2008; Uzzi, 1996).

Previous literature identifies several structural components in a network and the qualities of

BSRs as influential factors in the willingness to disclose information by supply chain members.

As a monumental study, Burt (1987) discussed how social structural circumstances make two

individuals assimilated into one another when two individuals maintain structurally equivalent

patterns in relation to all other individuals within the network. The identical pattern of relations

with occupants of other positions in the network helps to manage uncertainty between the two

actors. Also, structural equivalence between buying firms and suppliers uses the redundancy

benefits in that the same source of information that two organizations share allows for the benefits

of information redundancy (Burt, 1997). In other words, irrespective of how they feel about one

another, structurally equivalent organizations will exhibit similar thoughts and behaviors due to

their relationships with other actors in the network (Chae et al., 2020; Xiao et al., 2015; McPherson

et al., 2001; Friedkin, 1998). Furthermore, if two organizations’ relationships with other actors in

a network become increasingly similar, their sentiments of collaboration get stronger, boosting the

possibility of a link between them (Xiao et al., 2015; Burt, 1997).

In supply chains, the quality of BSRs and the structure of interorganizational networks are

key components of social capital, significantly shaping supply chain transparency, visibility, and

dependency (Melnyk et al., 2022; Gualandris et al., 2021; Cheung et al., 2020; Awaysheh and

Klassen, 2010). Direct BSRs promote the exchange of information and resources and facilitate the

diffusion of organizational practices (Pallotti and Lomi, 2011). Structurally equivalent actors, those

62

embedded in similar social circles, tend to engage in more consistent knowledge sharing and experi-

ence lower levels of conflict and perceived risk (Cannizzaro, 2020). Such homophilic relationships,

characterized by shared partners that reinforce prevailing norms and expectations, cultivate elevated

levels of trust and reciprocity, particularly as environmental standards and consumer expectations

diffuse through overlapping network ties. Consequently, in highly visible dyads marked by high

structural equivalence, the negative influence of a buyer on its supplier’s environmental disclosures

is attenuated.

In contrast, low structural equivalence, reflecting fewer overlapping connections,

limits visibility, increasing the likelihood that suppliers in arm’s length relationships will refrain

from disclosure to avoid reputational or commercial risks. Given that structurally equivalent actors

benefit from similar network patterns and visibility, we propose the following hypothesis:

Hypothesis 2A (𝐻2𝐴) Structural equivalence between a buyer and its supplier diminishes the

negative influence of the buyer on the supplier’s environmental disclosures.

3.3.2.2 The Influence of Social Capital’s Relational Dimension

The relational dimension of social capital that buying firms establish with their suppliers, on

the other hand, indicates the assets that are inherent in a relationship and are built based upon

mutual trust, respect, and friendship between two individuals through longstanding interactions

(Chae et al., 2020; Villena and Craighead, 2017; Krause et al., 2007; Nahapiet and Ghoshal, 1998;

Tsai and Ghoshal, 1998). Relational attributes in social networks, such as trust, trustworthiness,

cooperation, and relationship duration, help to reduce the uncertainty in relationships and to

create opportunities for the potential transactions (Squire et al., 2009; Uzzi and Lancaster, 2003;

Uzzi, 1996). Based upon relationship-specific communication and coordination routines, multiple

and repeated interactions could enhance cooperation and trust in the long run (Ravindran et al.,

2015; Krause et al., 2007; Hoetker, 2005; Uzzi, 1996), which, in turn, ends up with reducing

transactional uncertainty (Uzzi and Lancaster, 2003; Uzzi, 1996). Trust, as a pivotal element of

social interactions, serves primarily as a governance mechanism for managing relationships (Uzzi,

1996), motivates partners to collaborate (Tsai and Ghoshal, 1998), and enhances the efficacy of

information flows (Lee and Ha, 2018). Consequently, the risks associated with improper use

63

of information and information asymmetry among involved parties can be substantially reduced

through open and honest mutual information sharing, which is underpinned by trust (Lee and Ha,

2018).

The strength of ties has often been used as another representation of relational capital (Tate

et al., 2013; Krause et al., 2007; Hansen, 1999; Granovetter, 1973). To be specific, Granovetter

(1973) defines the strength of ties as “a combination of the amount of time, the emotional intensity,

the intimacy, and the reciprocal services which characterize the tie (Granovetter, 1973, p.1361).”

Rowley et al. (2000) also measures the strength of ties in two different ways: the frequency with

which partners contact and their level of commitment to the relationship in terms of resources.

In the context of SCM, previous studies have also conceptualized the strength of ties among

members from a network perspective. Kim (2014) considers relational embeddedness as the

strength of relationships between suppliers and buying firms, suggesting that a high degree of

relational embeddedness promotes close observations and interactions, enabling firms to access

specific information, tactical knowledge, and expertise. Krause et al. (2007) also highlights that

direct involvement in supplier development, defined as activities initiated by buyers to enhance

their suppliers’ performance, creates an environment conducive to the transfer of tacit knowledge

and fosters mutual learning between firms.

Relational embeddedness, defined by strong and enduring ties between actors, facilitates the

exchange of information and the assimilation of organizational practices (Andersson et al., 2002;

Uzzi, 1996). Relational capital helps mitigate opportunistic behavior by fostering trust and cooper-

ation, whereas its absence may lead partners to withhold valuable resources, reduce collaborative

behavior, and increase uncertainty (Villena et al., 2011). When firms maintain longstanding re-

lationships characterized by trust, shared vision, and high asset specificity, they are more likely

to share environmental information with supply chain partners (Kim and Henderson, 2015; Jira

and Toffel, 2013; Tate et al., 2011; Locke et al., 2007). Within our research context, this suggests

that in more visible dyads, characterized by longer relationship duration, the negative influence of

buyer environmental disclosure on supplier disclosure tends to be attenuated. Conversely, when

64

relationship duration is short and visibility is limited, suppliers may be more reluctant to disclose,

particularly in arm’s-length relationships. Given that the diffusion of sustainability practices, de-

fined as efforts to enhance a firm’s environmental responsibility, is closely tied to the relational

embeddedness of partners within BSRs (Jira and Toffel, 2013; Tate et al., 2011), we propose the

following hypothesis:

Hypothesis 2B (𝐻2𝐵) Relationship duration between a buyer and its supplier diminishes the

negative influence of the buyer on the supplier’s environmental disclosures.

3.3.2.3 The Influence of Social Capital’s Cognitive Dimension

Next, the cognitive dimension of social capital refers to shared languages or common cultures

that help social units to agree on collective goals and proper behaviors (Chae et al., 2020; Nahapiet

and Ghoshal, 1998; Tsai and Ghoshal, 1998). Corporations that align on common values are more

likely to form partnerships, facilitated by the exchange of resources (Blount and Li, 2021; Nahapiet

and Ghoshal, 1998; Tsai and Ghoshal, 1998). Social networks also keep evolving, facilitated

by various factors, such as shared activities, affiliations of involved parties, and the similarity of

individual attributes (Kossinets and Watts, 2006). As a foundational study, McPherson et al. (2001)

defines homophily as the principle that social units prefer significant contact with others who

share similar characteristics, resulting in more frequent interactions among similar entities. The

study also highlights the several causes of homophily. In their analysis, they identify geographical

location and organizational foci as primary motivators influencing the formation of homophilous

ties. The prevalence of homogeneity among network members, stemming from these fundamental

causes, is often considered a cognitive dimension of social capital, highlighting how the evolution

of social networks is predominantly driven by the similarity of attributes shared by organizations.

Geographical and cultural distances are known to hinder collaboration and the formation of

close relationships, making it difficult for firms to implement consistent monitoring and auditing

systems across diverse markets (Adhikary et al., 2020; Awaysheh and Klassen, 2010). Cultural

gaps between buyers and suppliers can disrupt cohesion, increase hold-up costs, and weaken col-

laboration and coordination. In contrast, BSRs are more likely to exhibit higher levels of cognitive

65

social capital when both parties share similar political, economic, and cultural backgrounds (Chae

et al., 2020; Bolino et al., 2002). In other words, shared values, such as institutional ownership and

domain similarity, help connect entities within a network, facilitating task completion and fostering

cooperation (Cheung et al., 2020; Van de Ven, 1976). Their connections can also be strengthened by

shared physical time and space. Thus, while firms are obligated to disclose their environmental per-

formance to effectively manage inter-firm environmental impacts, their responses to environmental

reporting requirements vary based on situational factors. These factors are often determined by

whether organizations share similar behaviors and attitudes, enabling them to actively participate

in sustaining inter-organizational relationships and enhancing transparency in supply chains.

Hence, we expect that the cognitive dimension of social capital, cultural similarity, moderates the

negative influence of a buyer on its supplier’s environmental disclosures. Specifically, a supplier’s

disclosure tends to be lower when supply chain visibility is reduced due to low cultural similarity

with the buyer. In such cases, the negative relationship is likely to be more pronounced, as suppliers

in culturally distant, arm’s length relationships may be more reluctant to disclose, possibly to avoid

commercial or reputational risks. Conversely, in more visible dyads, characterized by high cultural

similarity, the negative effect is attenuated. Taking these considerations into account, we further

posit that:

Hypothesis 2C (𝐻2𝐶) National cultural similarity between a buyer and its supplier diminishes

the negative influence of the buyer on the supplier’s environmental disclosures.

3.3.3 Conceptual Framework for Hypothesis Testing

In this study, we formulate hypotheses to test the direct influence (𝐻1) of a buyer on its supplier

environmental disclosures. Figure 3.1 illustrates the conceptual framework, depicting the proposed

direct influence between the supplier environmental disclosure and that of a buying firm. It also

shows the moderating effects of different dimensions of social capital embedded in BSRs (resp.,

the structural dimension of social capital (𝐻2𝐴), the relational dimension of social capital (𝐻2𝐵),

and the cognitive dimension of social capital (𝐻2𝐶)). In total, we test four hypotheses in this study.

66

Figure 3.1 Conceptual Framework

Structural Dimension
of Social Capital
in Dyads

Relational Dimension
of Social Capital
in Dyads

Cognitive Dimension
of Social Capital
in Dyads

𝐻2𝐴 (+)

𝐻2𝐵 (+)

𝐻2𝐶 (+)

𝐻1 (−)

Buyer
Environmental
Disclosure

Supplier
Environmental
Disclosure

3.4 Data Source

3.4.1 Data

In this section, we first outline the multiple sources for collecting secondary data. Subsequently,

we detail the variables utilized in our empirical models to test our hypotheses.

In this study, to establish BSRs, we use the dataset that comes from FactSet Revere. The FactSet

Revere supply chain relationships data illuminates the interconnected commercial partnerships

between organizations. Data about supply chain relationships provided by FactSet Revere is

systemically collected through primary public sources, including U.S. Securities and Exchange

Commission (SEC) 10-K annual reports, public announcements, and investor relations, and reverse-

linked to non-disclosure parties (FactSet Revere, 2021). Regarding the scope of data, the FactSet

supply chain relationships cover over 31,000 publicly traded companies globally, encompassing

more than 450,000 business relationships (FactSet Revere, 2021). According to FactSet Revere, the

collected supply chain relationships are predominantly categorized into four main types, resulting in

13 distinct sub-types of supply chain relationships to address varying levels of company disclosure.

Specifically, inter-company relationships are organized and presented based on the nature of the

relationship, competitors, strategic partners, suppliers, and customers, each further divided into 13

normalized sub-types for more precise categorization.

The FactSet Revere supply chain relationships data specifically includes unique data for this

study in that it also provides information on the start date and end date of the business relationships

and interconnections, with the history of data going all the way back to 2003. In addition to these

data, the nature of a company’s relationships is further detailed beyond the four categories and

67

thirteen sub-types, and is summarized as relationship keywords within the dataset. In conclusion,

FactSet Revere supply chain relationships data gives 360-degree visibility into a company’s in-

terconnected business ties, their nature, and their dependability. To operationalize social capital

factors and to form supply networks of focal firms, FactSet Revere supply chain relationships data

is primarily used in this study.

To examine the fundamental relationship between buyer and supplier environmental disclosures,

we employ Bloomberg’s ESG disclosure score from the Bloomberg ESG database, which compiles

firm-level quantitative and policy-related environmental indicators from both public and private

sources, as our primary measure of environmental disclosure (Diebel et al., 2024; Bellamy et al.,

2020). Specifically, the database collects data on firms’ internal ESG practices and performance

through direct communications, such as meetings, phone interviews, and surveys, as well as from

corporate sustainability reports, regulatory filings, websites, and news articles (Gualandris et al.,

2021). Bloomberg’s ESG data provides comprehensive data on over 9,000 companies from more

than 70 countries, spanning over 900 fields related to ESG topics like air quality and governance,

and offers up to a decade of historical data that supports quantitative analysis to identify ESG factors

that may impact long-term company performance (Bloomberg Professional Services, 2018).

The Bloomberg ESG disclosure score provides investors with insights into how transparently

companies report their ESG practices. It measures the quality of ESG disclosures based on over

120 data points across environmental, social, and governance factors but does not assess the actual

performance of these practices. A firm’s performance across each indicator is validated through

multiple sources, including CSR reports, annual reports, company websites, Carbon Disclosure

Project (CDP) data, and third-party research to guarantee accuracy and reliability (Bellamy et al.,

2020; Bloomberg Professional Services, 2018). Hence, to operationalize a firm’s extent of en-

vironmental disclosure, we utilize Bloomberg’s environmental disclosure score, which quantifies

the level of a firm’s environmental reporting and assigns firms a numerical value based on the

comprehensiveness of their environmental disclosures (Bellamy et al., 2020).

Third, we compile firm financial data from the Bloomberg Financial Analysis (FA) database

68

(Diebel et al., 2024; Gualandris et al., 2021; Bellamy et al., 2020; Sharma et al., 2020). This

database serves as an extensive repository of financial data, offering both real-time and historical

information on public and private companies. It encompasses market data, economic indicators, and

sector-specific metrics, providing a holistic view of financial landscapes. To construct independent

and control variables, we also augment our three primary data sources with additional private

datasets. These include the base culture data encompassing Geert Hofstede’s six dimensions of

culture, and FactSet Revere’s Company data, which provides general information about companies.

3.4.2 Sample Collection Procedure

In this section, we first outline the composition of our sample for empirical analysis. Following

that, we elaborate on the procedure used to collect our dataset and proceed to describe the variables

integrated into our empirical models for hypothesis testing.

3.4.2.1 Sample Selection

In this study, we hypothesize the impact of buyer environmental disclosure on supplier en-

vironmental disclosure. Accordingly, our primary unit of analysis is the supplier within a BSR.

Since our objective is to investigate the primary relationship and the moderating effects from a

network perspective, we have framed our sample, which encompasses a diverse array of buyers,

to initiate our data collection. By referring to previous literature (Diebel et al., 2024; Gualandris

et al., 2021), the Forbes 2000 list published in 2018 serves as our fundamental sample framework in

this study because it offers an annual ranking of the world’s largest firms across various sectors in

both manufacturing and service industries. These firms are equally visible to external stakeholder

groups, a factor crucial due to their significant business volumes (Gualandris et al., 2021).

In our industry selection process, we specifically focus on buying firms within the broader

category of manufacturing. This choice is motivated by the fact that traditional manufacturing

industries are significant contributors to negative environmental externalities (Hardcopf et al.,

2021). In previous research, many papers focused on SCM tend to examine a single industry where

buyers operate, such as the automotive (Chae et al., 2020; Potter and Wilhelm, 2020; Kim et al.,

2011), electronics (Chedid et al., 2021; Wilhelm and Villena, 2021; Basole et al., 2017), and food

69

industries (Bourlakis et al., 2014; Grimm et al., 2014). Conversely, some studies simultaneously

explore multiple industries to facilitate comparisons (Diebel et al., 2024; Adhikary et al., 2020;

Bellamy et al., 2020; Hardcopf et al., 2019; Villena, 2019; Villena and Gioia, 2018; Wilhelm

et al., 2016). In this study, based on the Global Industry Classification Standard (GICS) industry

classification, we examine the supply networks of buying firms between two distinct manufacturing

sectors: Materials (four-digit GICS code: 1510) and Pharmaceuticals & Biotechnology industries

(four-digit GICS code: 3520). In previous literature, the materials industry has often been classified

as environmentally sensitive (see, for example, Cho et al. (2012), Reverte (2009), Cho and Patten

(2007), Patten (2002), and Cowen et al. (1987)). Given this context, the industry is likely more

attuned to concerns regarding safety, environmental impact, and sustainability.

In contrast, the

pharmaceutical industry is recognized for its significant investment in research and development

(R&D). However, with an aging global population and advancements in healthcare systems fueling

demand, this industry faces increasing pressure to achieve sustainable development while expanding

its product offerings. Consequently, the environmental impact of its operations has become a focal

point of scrutiny (Milanesi et al., 2020). To enrich our investigation, this study proposes examining

two distinct manufacturing industries with respect to potential environmental stewardship. This

approach provides a more substantial justification for focusing on specific sectors of interest.

In the sample collection procedure shown in Figure 3.2, we exclusively use supply chain

relationships classified as “direct” (disclosed by the focal firm) and “reverse” (disclosed by the

focal firm’s suppliers or customers). This approach enables us to construct a wide range of dyadic

relationships across two distinct manufacturing industries (Culot et al., 2023). After organizing our

data, we integrate this sample with other datasets described in Section 3.4.1. We then eliminate

dyads lacking key independent and control variables in two steps: first, we exclude dyads with

missing values for the moderators, resulting in 10,145 dyads, 1,608 unique suppliers, and 166

unique buyers; second, we remove dyads missing control variable data, involving 9,296 dyads,

1,459 unique suppliers, and 166 unique buyers.

70

Figure 3.2 Sample Selection Procedure

Step 1: Select Buyers

1. Download the 2018 Forbes Global 2000 firm list.

2. Select 208 manufacturing firms in Materials (GICS: 1510) and Pharmaceuticals (GICS: 3520).

Step 2: Construct Buyer-supplier Dyads

1. Download FactSet Revere BSRs data (2015–2019) for relationships identified as customers or suppliers,
excluding dyads missing International Securities Identification Number (ISIN) codes for both buyers
and suppliers (857,372 dyads).

2. Align the selected BSRs with buying firms featured in the 2018 Forbes Global 2000 rankings,

specifically focusing on the two aforementioned distinct industries, both of which are included in the
general category of manufacturing (32,413 dyads).

3. Consolidate the data by merging dyads according to relationship IDs and respective years, ensuring
each unique ID corresponded to a single BSR within each year from 2015 to 2019 (23,416 dyads).

4. Exclude dyads that lack information on the suppliers’ GICS sector and industry codes, as well as those

classified under the service industry (18,040 dyads & 5,914 unique dyads).

Step 3: Merge Environmental Disclosure Data

1. Collect environmental disclosure data for buyers and suppliers for the period from 2015 to 2019.

2. Exclude dyads with missing environmental disclosure data for buyers and suppliers 10,580 dyads &

1,664 unique suppliers & 169 unique buyers).

Step 4: Form Supply Networks of Buying Companies

1. Separate lists of buyers and suppliers were created for each year within the period 2015 to 2019.

2. Remove duplicates, resulting in 857,872 buyer relationships and 356,824 supplier relationships.

3. Based on Step 4-2, retain any BSRs that were active at any point from 2015 to 2019.

4. Based on buyers and suppliers identified in Step 3-2, 30,117 (2015), 36,221 (2016), 43,481 (2017),
44,778 (2018), and 46,399 (2019) dyads were retained; structural equivalence was then computed.

3.4.2.2 Sample Description

Previous research has highlighted the crucial role of partner interconnectedness in enhancing

information flows within a network (Bellamy et al., 2020). To better depict the unique structural

features of supply networks across the chosen industries in 2019, as outlined in the fourth point of

Step 2 in Figure 3.2, we employ Gephi 0.10.1. As one of the network mapping layouts provided by

this visualization tool, we employ the Yifan Hu multilevel layout. This layout method integrates

71

a force-directed model with a graph coarsening technique, effectively simplifying complex visual-

izations while preserving the integrity of the network. This approach is efficiently detailed by Hu

(2005), highlighting its effectiveness in network analysis.

(a) Materials Industry in 2019

(b) Pharmaceutical Industry in 2019

Figure 3.3 Visualization of Supply Networks of Two Different Industries

Figure 3.3 clearly shows that the materials industry was denser compared to the pharmaceutical

industry in 2019.

In addition, Figures 3.3a and 3.3b use color-coded modularity to delineate

firms into distinct communities, facilitating an intuitive understanding of network segmentation.

Based on the color-coded modularity for community detection, the materials industry comprised 17

communities, while the pharmaceutical industry included 15 in 2019. Furthermore, the materials

industry maintained a larger supply base compared to the pharmaceutical industry during the same

period.

In summary, our final sample includes suppliers and buyers operating across diverse manu-

facturing sectors (see Table 3.2). Additionally, based on the location of the headquarters of the

suppliers in our sample, Figure 3.4 demonstrates the global coverage of these suppliers. The num-

bers shown in Figure 3.4 represent the observed number of suppliers according to the location of

their headquarters in Table 3.2.

3.4.3 Variables

In this section, we construct a set of variables for empirical analysis and hypothesis testing.

These variables are detailed in Table 3.4, which presents their descriptions and applications in

72

Table 3.2 Sample Demographics

Unique Suppliers

Unique Buyers

Industry

Automobiles & Components

Capital Goods

Consumer Durables & Apparel

Energy

Food, Beverage & Tobacco

Health Care Equipment

Household & Personal Products

Materials

Pharmaceuticals & Biotechnology

Semiconductors & Equipment

Technology Hardware

Utilities

Total

Industry

Materials

Pharmaceuticals & Biotechnology

N

128

38

N

27

316

17

130

21

56

6

420

321

20

43

82

1,459

Total

166

Country of Suppliers’ Headquarters in Unique Dyads

Country of Buyers’ Headquarters in Unique Dyads

Country

United States

India

Japan

China

Australia

United Kingdom

Germany

South Korea

France

Canada

Others

Total

N

1,161

316

315

186

177

166

157

149

111

97

804

3,639

Country

United States

Japan

Germany

United Kingdom

South Korea

France

India

Australia

Canada

Luxembourg

Others

Total

N

1,016

458

335

238

160

156

156

143

118

115

744

3,639

the analysis.

In addition, Table 3.3 provides the summary statistics and correlation matrices,

which are crucial for understanding the interactions among the variables relevant to our empirical

investigation.

73

Figure 3.4 The Distribution of Suppliers’ Headquarters across the World

3.4.3.1 Dependent Variable

As a primary dependent variable, we utilize the voluntary environmental disclosure of supplier 𝑠,

denoted as 𝑆_𝐸 𝐷 𝑠,𝑡+1, which is rated on a 100-point scale, for the subsequent year 𝑡 + 1 in this study.

This rating draws upon 120 internal and external environmental impact indicators, each weighted

according to the salience of industry-specific characteristics, such as emissions, energy use, water

management, waste management, and other environmental operational categories (Diebel et al.,

2024; Bellamy et al., 2020). Consistent with Diebel et al. (2024), Villena and Dhanorkar (2020), and

Jira and Toffel (2013), this study explores the impact of buyers’ sustainability attributes on the en-

vironmental transparency of suppliers. One key strength of Bloomberg’s environmental disclosure

score from the Bloomberg ESG database is its incorporation of both structured and unstructured

data, with Bloomberg’s Global Reporting Initiative (GRI)-based sustainability survey filling gaps

in publicly available information (Bellamy et al., 2020). Therefore, Bloomberg’s environmental

disclosure score is ideal for our study, as it specifically measures a supplier environmental disclosure

rather than actual environmental performance, while also accounting for the firm’s industry context

(Bellamy et al., 2020; Yu et al., 2018). Previous literature has also utilized Bloomberg’s ESG

74

disclosure score to assess firms’ willingness to participate in sustainability activities and practices

(Diebel et al., 2024; Adhikary et al., 2020; Bellamy et al., 2020).

3.4.3.2

Independent Variables

As the primary independent variable, we employ the voluntary environmental disclosure of the

buyer 𝑏, denoted as 𝐵_𝐸 𝐷 𝑏,𝑡, which is rated on a 100-point scale, for the current period 𝑡 in this

study. Consistent with the dependent variable, we utilize Bloomberg’s environmental disclosure

score from the Bloomberg ESG database.

In addition, we utilize the three dimensions of social capital to examine the moderating effects

of a buyer’s influence on supplier environmental disclosure (resp., structural dimension of social

capital (i.e., 𝑆𝐷 𝑠,𝑏,𝑡), relational dimension of social capital (i.e., 𝑅𝐷 𝑠,𝑏,𝑡), and cognitive dimension

of social capital (i.e., 𝐶𝐷 𝑠,𝑏,𝑡)). To more accurately evaluate the impact of the structural dimension

of social capital, denoted as 𝑆𝐷 𝑠,𝑏,𝑡, on the dyadic relationships between buyers and suppliers, it is

crucial to examine structural equivalence from both parties’ perspectives. This involves examining

the patterns and configurations of relationships each party maintains within the network, thus

offering insights into how similar their roles and influences are within their respective networks.

Structural equivalence indicates how similar relations between two nodes in a network are (Xiao

et al., 2015; Burt, 1997; Coleman, 1994). In a few previous studies, structural equivalence has

been operationalized in a similar manner by assessing the shared connections of two separate nodes

(Chae et al., 2020; Xiao et al., 2015; Kossinets and Watts, 2006). In this research context, structural

equivalence will be measured by the proportion of shared suppliers or customers between the two

firms in a BSR, relative to their total number of connections. This measure, commonly referred

to as the Jaccard index, calculates the ratio of shared (or overlapping) connections to the total

connections, reflecting the intersection of true and expected positive matches (Ebbes and Netzer,

2021). We compute this metric in the following manner:

75

𝐽 (𝐵𝑖, 𝑆 𝑗) =

|𝑁 (𝐵𝑖) ∩ 𝑁 (𝑆 𝑗)|
|𝑁 (𝐵𝑖) ∪ 𝑁 (𝑆 𝑗)|

where 𝑁 (𝐵𝑖) is the neighborhood nodes of 𝐵𝑖,

(3.1)

𝑁 (𝑆 𝑗) is the neighborhood nodes of 𝑆 𝑗 .

In this study, we calculate 𝐽 (𝐵𝑖, 𝑆 𝑗 ) as defined in (3.1) using the Python NetworkX package,

which is widely utilized in supply network research (Chae et al., 2022; Taghizadeh et al., 2021).

Identifying these shared connections is crucial for assessing the structural equivalence between

firms.

The relational dimension of social capital, denoted as 𝑅𝐷 𝑠,𝑏,𝑡, which represents the second

dimension of social capital, will be operationalized through the extent to which both the buyer and

the supplier hinge on their relationship. In line with the previous work (Chae et al., 2020; Hoetker

et al., 2007), the duration (in years) of BSRs is measured by the total number of years a buying firm

has maintained relationships with its suppliers. In a recent and important advancement, Capaldo

(2007) expands on prior relationship theories by conceptualizing inter-organizational relationship

strength as a three-dimensional construct encompassing temporal, resource, and social dimensions.

In the study, Capaldo (2007) identifies three key factors that interact to determine the strength of BSR

ties: (1) the overall duration of the relationship, (2) the frequency of interaction, and (3) the intensity

of interaction. In this framework, duration represents the temporal dimension, while all three factors

collectively embody the resource-based and social dimensions. In essence, higher levels of these

factors generally correspond to greater resource commitments and are prerequisites for the transfer

of social content between firms (Autry and Golicic, 2010). Hence, the operationalization of the

relational dimension of social capital is well justified.

To effectively measure the cultural gap between a buying firm and its suppliers, previous

research has frequently employed the concept of national cultural distance (Chae et al., 2020;

Morosini et al., 1998). In this study, we adapt the traditional national cultural distance measure

to assess the cognitive dimension of social capital, denoted as 𝐶𝐷 𝑠,𝑏,𝑡. By reverse engineering

this metric, we aim to capture the shared understanding evident in language, codes, and narratives

76

(Chae et al., 2020; Claridge, 2018; Morosini et al., 1998; Kogut and Singh, 1988). The measure of

national cultural distance reflects the Euclidean distance between Hofstede’s cultural dimensions

for the countries where the buyer’s and supplier’s headquarters are located. Specifically, following

the established framework of Beugelsdijk et al. (2018) and Hofstede (2011), this study incorporates

six cultural dimensions: power distance, uncertainty avoidance, individualism versus collectivism,

masculinity versus femininity, long-term versus short-term orientation, and indulgence versus

restraint. To capture the shared understanding reflected in these cultural dimensions, we multiply

it by a negative one to refine it into a more context-specific proxy for cognitive alignment or

misalignment in international business relationships. Specifically, we compute this metric in the

following manner:

𝐶𝐷 𝑠,𝑏,𝑡 = −1 × (The Euclidean distance between Hofstede’s cultural dimensions) .

(3.2)

3.4.3.3 Control Variables

To avoid the confounding effects stemming from unobserved dyad and supplier heterogeneity,

we utilize several factors that can address supplier-level and dyad-level heterogeneity in this study

and might affect a supplier’s level of environmental disclosure.

Corporate governance is primarily categorized into internal mechanisms, such as the board

of directors, and external mechanisms, including ownership and pressure from other stakeholders

(Velte, 2024). Given that corporate environmental disclosure is considered a high-level strategic

decision for the firm, we control for the number of board members of supplier 𝑠 in year 𝑡 (i.e.,

𝑆_𝐵𝑂 𝐴𝑅𝐷_𝑆𝐼 𝑍 𝐸𝑠,𝑡) (Velte, 2024; Lu and Shang, 2017). Furthermore, as a key aspect of corporate

governance, we control for gender diversity at the executive level, measured by the number of female

executives within a company (i.e., 𝑆_𝐹𝐸 𝑀_𝐸 𝑋 𝐸𝐶𝑈𝑇 𝐼𝑉 𝐸 𝑆𝑠,𝑡). Recognized as a crucial variable

in sustainable corporate governance, gender diversity is predicated on the idea that female directors

contribute diverse perspectives, promote a broader array of eco-friendly solutions, and influence

the strategic and operational decisions of boards toward addressing environmental issues (Velte,

2024; Liu, 2018). Existing literature on SSCM has demonstrated that the gender composition

77

of the workforce can significantly impact sustainability improvements, as well as environmental

and social conditions (Villena et al., 2021; Wilhelm and Villena, 2021). Therefore, justifying the

inclusion of the number of female executives within a company as a control variable in this study

is well-supported.

A firm’s financial performance may affect its willingness and ability to disclose environ-

mental information (Bellamy et al., 2020). To account for this influence, we control for a fi-

nancial metric that reflects firm financial performance, including net income after taxes (i.e.,

𝑆_𝑁 𝐸𝑇_𝐼 𝑁𝐶𝑂 𝑀 𝐸𝑠,𝑡). Next, we also control for supplier size, measured as the natural logarithm

of the number of employees (i.e., ln 𝑆_𝐸 𝑀 𝑃𝑠,𝑡) (Villena et al., 2021; Wilhelm and Villena, 2021;

Adhikary et al., 2020; Sharma et al., 2020), given its role as a well-established determinant of

corporate environmental disclosure (D’Amico et al., 2016; Cho et al., 2012; Cormier and Magnan,

1999).

We incorporate ESG and socio-political factors to account for CSR activities. First, the pres-

ence of a dedicated committee (i.e., 𝑆_𝐶𝑆𝑅_𝐶𝑂 𝑀𝑠,𝑡) overseeing CSR and sustainability initiatives

offers key insights into a company’s governance structure and commitment to sustainability (Velte,

2024). Another key environmental factor is a company’s exposure to climate-related risks (i.e.,

𝑆_𝐶 𝐿𝐼_𝑅𝐼𝑆𝐾𝑆𝑠,𝑡), particularly whether it has identified inherent risks with the potential to sig-

nificantly impact its financial or strategic operations. Corporate climate risks, goals, and related

information are typically disclosed through environmental reporting (Wilhelm and Villena, 2021;

Jira and Toffel, 2013). Moreover, firms often prioritize accountability to their workforce to gain or

maintain corporate legitimacy, recognizing employees as key stakeholders (Parsa et al., 2018). As

companies navigate public pressure within social and political environments, they strengthen their

legitimacy when stakeholders perceive their value systems as acceptable. This social legitimacy is

continuously assessed and reinforced through the public policy process (Parsa et al., 2018; Patten,

2002). In this regard, we also account for the presence of companies’ human rights policies (i.e.,

𝑆_𝐻𝑅_𝑃𝑂 𝐿𝐼𝐶𝑌𝑠,𝑡) as a control variable.

Lastly, organizations actively adapt to their evolving environments and contingency factors, such

78

as technological advancements and government policies, to optimize performance (Sodhi and Tang,

2018). In OM, variables such as national context, culture, firm size, and strategic orientation are

commonly recognized as contingencies shaping corporate best practices. In SSCM, in particular,

industry-specific and national contexts serve as key contingency factors (Sodhi and Tang, 2018;

Sousa and Voss, 2008). Within this framework, environmental disclosure in supply chains reflects

a company’s commitment to complying with environmental regulations and norms. However, these

commitments may be compromised not only in the countries where suppliers operate but also in the

buyer’s market (Sodhi and Tang, 2019). Previous studies highlight industry type, country context,

and peer effects as major contingency factors influencing the extent of corporate disclosure (Zhao

and Wang, 2024; Seo, 2021; Bellamy et al., 2020; Sodhi and Tang, 2018; Jira and Toffel, 2013;

Cowen et al., 1987).

In this study, due to the high correlation between the geographical proximity in BSRs and one

of our independent variables, 𝐶𝐷 𝑠,𝑏,𝑡, we have opted to exclude the variable indicating whether the

buying firm and supplier are headquartered in the same country from the main analysis. However,

we retain this variable for robustness checks as an alternative metric for assessing the cognitive

dimension of social capital. In addition, to control for industry alignment at the dyadic level, we

include a dummy variable, 𝑆 𝐴𝑀 𝐸_𝐼 𝑁 𝐷 𝑠,𝑏,𝑡, which denotes whether the buying firm and supplier

operate within the same industry, as determined by four-digit GICS industry classifications (Song

et al., 2023). This methodological choice aids in demonstrating the consistency between our main

findings and the robustness check results. The summary statistics and correlation matrix for the

variables of interest are presented in Table 3.3.

3.5 Empirical Specification

In this section, we outline the models used for the empirical analysis to test our hypotheses and

conclude with a presentation of the empirical results.

3.5.1 Empirical Models

To mitigate causal ambiguity, we measure each supplier’s environmental disclosure at time t+1

(𝑆_𝐸 𝐷 𝑠,𝑡+1), while the buyer’s disclosure score (𝐵_𝐸 𝐷 𝑏,𝑡), three social capital factors (𝑆𝐷 𝑠,𝑏,𝑡,

79

Table 3.3 Descriptive Statistics and Correlations

SD

0
0
0
0

N Mean

Min Max 1

2

3

4

5

6

7

90.577 1.00
90.577 0.03* 1.00
0.394 0.23* -0.00 1.00

Variable
1. 𝑆_𝐸 𝐷 𝑠,𝑡+1
5,571 27.187 24.857
2. 𝐵_𝐸 𝐷 𝑏,𝑡
9,296 52.819 19.347
3. 𝑆𝐷 𝑠,𝑏,𝑡
0.027
9,296 0.021
4. 𝑅𝐷 𝑠,𝑏,𝑡
9,296 2.023
2.110
5. 𝐶𝐷 𝑠,𝑏,𝑡
9,296 -45.836 38.061 -129.171
6. 𝑆_𝐵𝑂 𝐴𝑅𝐷_𝑆𝐼 𝑍 𝐸𝑠,𝑡
9,296 9.271
7. 𝑆_𝐹𝐸 𝑀_𝐸 𝑋 𝐸𝐶𝑈𝑇 𝐼𝑉 𝐸 𝑆𝑠,𝑡 9,296 0.909
8. 𝑆_𝑁 𝐸𝑇_𝐼 𝑁𝐶𝑂 𝑀 𝐸𝑠,𝑡
9. ln 𝑆_𝐸 𝑀 𝑃𝑠,𝑡
10. 𝑆_𝐶𝑆𝑅_𝐶𝑂 𝑀𝑠,𝑡
11. 𝑆_𝐶 𝐿𝐼_𝑅𝐼𝑆𝐾𝑆𝑠,𝑡
12. 𝑆_𝐻𝑅_𝑃𝑂 𝐿𝐼𝐶𝑌𝑠,𝑡
13. 𝑆 𝐴𝑀 𝐸_𝐼 𝑁 𝐷 𝑠,𝑏,𝑡

16 0.03 -0.01 0.10* 1.00
0
-0.16* -0.10* 0.13* 0.06* 1.00
22 0.39* -0.02 0.20* -0.02* -0.02* 1.00
18 0.23* 0.03* 0.14* 0.05* -0.08* 0.22* 1.00
9,296 580.128 2376.88 -22355 23352 0.30* -0.02 0.19* 0.05* -0.06* 0.22* 0.17*
13.348 0.63* -0.02 0.18* -0.02* -0.13* 0.54* 0.20*
9,296 8.322
0.24* 0.00 0.11* -0.01 -0.00 0.30* 0.19*
9,296 0.300
0.29* 0.05* 0.05* 0.10* -0.02 0.24* 0.18*
9,296 0.221
0.65* 0.05* 0.17* -0.02 -0.14* 0.31* 0.11*
9,296 0.610
9,296 0.512
0.02 -0.09* 0.18* 0.12* -0.08* -0.04* -0.00
N Mean

2.027
0.458
0.415
0.488
0.500
SD

Min Max 8

3.020
1.212

0
0
0
0
0

1
1
1
1

3
0

11

10

12

13

9

8. 𝑆_𝑁 𝐸𝑇_𝐼 𝑁𝐶𝑂 𝑀 𝐸𝑠,𝑡
9. ln 𝑆_𝐸 𝑀 𝑃𝑠,𝑡
10. 𝑆_𝐶𝑆𝑅_𝐶𝑂 𝑀𝑠,𝑡
11. 𝑆_𝐶 𝐿𝐼_𝑅𝐼𝑆𝐾𝑆𝑠,𝑡
12. 𝑆_𝐻𝑅_𝑃𝑂 𝐿𝐼𝐶𝑌𝑠,𝑡
13. 𝑆 𝐴𝑀 𝐸_𝐼 𝑁 𝐷 𝑠,𝑏,𝑡
Note: ∗𝑝 < 0.05.

9,296 580.128 2376.88 -22355 23352 1.00
9,296 8.322
9,296 0.300
9,296 0.221
9,296 0.610
9,296 0.512

2.027
0.458
0.415
0.488
0.500

0
0
0
0
0

1
1
1
1

13.348 0.29* 1.00

0.17* 0.28* 1.00
0.15* 0.30* 0.23* 1.00
0.17* 0.53* 0.25* 0.19* 1.00
0.05* -0.18* -0.05* -0.03* -0.09* 1.00

𝑅𝐷 𝑠,𝑏,𝑡, and 𝐶𝐷 𝑠,𝑏,𝑡), the dummy variable (𝑆 𝐴𝑀 𝐸_𝐼 𝑁 𝐷 𝑠,𝑏,𝑡) indicating whether buyers and sup-

pliers share the same institutional context, and all other covariates (𝝁𝒌,𝒕) are measured at time t

(Diebel et al., 2024; Dai et al., 2021). As shown in Table 3.5, 𝑭𝑬𝒕 represents the fixed effects

incorporated in the model, including supplier-country, supplier-industry, and year.

𝑆_𝐸 𝐷 𝑠,𝑡+1 = 𝛽0 + 𝛽1𝐵_𝐸 𝐷 𝑏,𝑡 + 𝛽2𝑆𝐷 𝑠,𝑏,𝑡

(cid:0)or 𝑅𝐷 𝑠,𝑏,𝑡, or 𝐶𝐷 𝑠,𝑏,𝑡 (cid:1)

+ 𝛽3𝑆𝐷 𝑠,𝑏,𝑡

(cid:0)or 𝑅𝐷 𝑠,𝑏,𝑡, or 𝐶𝐷 𝑠,𝑏,𝑡 (cid:1) × 𝐵_𝐸 𝐷 𝑏,𝑡

+

𝐾
∑︁

𝑘=1

𝛽𝑘 𝝁𝒌,𝒕 + 𝑭𝑬𝒕 + 𝜖𝑡+1.

(3.3)

We utilize ordinary least squares (OLS) estimation with multi-way clustered standard errors

at both the supplier-year and buyer-year levels to account for potential cross-sectional correlations

among dyads that share the same suppliers and buyers within each period (Diebel et al., 2024;

Cameron et al., 2011). We conduct our estimation using the 𝑟𝑒𝑔ℎ𝑑𝑓 𝑒 command in Stata 17, which

efficiently handles high-dimensional fixed effects and multi-way clustering.

80

Table 3.4 Variable Descriptions

Variable

Description

Dependent Variable
𝑆_𝐸 𝐷 𝑠,𝑡+1

Independent Variables
𝐵_𝐸 𝐷 𝑏,𝑡

𝑆𝐷 𝑠,𝑏,𝑡

𝑅𝐷 𝑠,𝑏,𝑡

𝐶𝐷 𝑠,𝑏,𝑡

Supplier’s environmental disclosure score (100-point scale), as-
sessing voluntary environmental reporting.
Reference: Diebel et al. (2024)

Buyer’s environmental disclosure score (100-point scale), mea-
suring sustainability transparency.
Reference: Diebel et al. (2024)
Structural dimension of social capital, measured using the Jaccard
index of shared supplier/buyer ties.
Reference: Xiao et al. (2015); Burt (1997)
Relational dimension, measured as the total duration of the buyer-
supplier relationship.
Reference: Capaldo (2007); Hoetker et al. (2007)
Cognitive dimension, measured as the negative value of national
cultural distance using Hofstede’s index.
Reference: Beugelsdijk et al. (2018); Hofstede (2011)

Control Variables
𝑆_𝐵𝑂 𝐴𝑅𝐷_𝑆𝐼 𝑍 𝐸𝑠,𝑡

Number of board members in the supplier firm.
Reference: Velte (2024); Lu and Shang (2017)

𝑆_𝐹𝐸 𝑀_𝐸 𝑋 𝐸𝐶𝑈𝑇 𝐼𝑉 𝐸 𝑆𝑠,𝑡 Number of female executives in the supplier firm.

𝑆_𝑁 𝐸𝑇_𝐼 𝑁𝐶𝑂 𝑀 𝐸𝑠,𝑡

ln 𝑆_𝐸 𝑀 𝑃𝑠,𝑡

𝑆_𝐶𝑆𝑅_𝐶𝑂 𝑀𝑠,𝑡

𝑆_𝐶 𝐿𝐼_𝑅𝐼𝑆𝐾𝑆𝑠,𝑡

𝑆_𝐻𝑅_𝑃𝑂 𝐿𝐼𝐶𝑌𝑠,𝑡

𝑆 𝐴𝑀 𝐸_𝐼 𝑁 𝐷 𝑠,𝑏,𝑡

Reference: Velte (2024); Liu (2018)
Supplier’s net profit (losses) after expenses.
Reference: Pekovic et al. (2018)
Natural log of the number of employees.
Reference: Villena et al. (2021); Wilhelm and Villena (2021)
Indicator for CSR or sustainability committee.
Reference: Velte (2024)
Indicator for substantive climate-related risk disclosure.
Reference: Wilhelm and Villena (2021); Jira and Toffel (2013)
Indicator for formal human rights policy.
Reference: Parsa et al. (2018)
Dummy for buyer and supplier in the same 4-digit GICS industry.
Reference: Song et al. (2023)

Database

Bloomberg

Bloomberg

FactSet

FactSet

Hofstede Index

Bloomberg

Bloomberg

Bloomberg

Bloomberg

Bloomberg

Bloomberg

Bloomberg

FactSet

3.5.2 Empirical Results

We present the results of the panel regression model in Table 3.5, estimating six different models

across two industries: materials and pharmaceuticals. First, we report the baseline model (Model

1), which includes only control variables. Then, we sequentially introduce the main independent

variables in Models 2 through 5. Finally, Model 6 presents the full specification, incorporating all

81

independent and control variables.

Table 3.5 Estimates of Ordinary Least Square Methods

Model 1

Model 2

Model 3

Model 4

Model 5

Model 6

-0.010
(0.014)

0.013
(0.017)

-0.017
(0.012)

-0.033*
(0.014)
2.638
(17.100)
0.722**
(0.273)

Variable

𝐵_𝐸 𝐷 𝑏,𝑡

𝑆𝐷 𝑠,𝑏,𝑡

𝐵_𝐸 𝐷 × 𝑆𝐷

𝑅𝐷 𝑠,𝑏,𝑡

𝐵_𝐸 𝐷 × 𝑅𝐷

𝐶𝐷 𝑠,𝑏,𝑡

𝐵_𝐸 𝐷 × 𝐶𝐷

0.006
(0.021)
13.797
(17.452)
0.596*
(0.279)
0.289+
(0.172)
-0.005
(0.004)
-0.054**
(0.017)
0.001*
(0.000)
0.377*
(0.179)
0.419
(0.311)
0.001***
(0.000)
4.057***
(0.272)
5.652***
(1.143)
4.352***
(1.029)
16.632***
(1.043)
-23.988***
(2.203)

Yes
Yes
Yes
Yes
5,570
0.560

0.209
(0.181)
-0.003
(0.004)

0.385*
(0.179)
0.477
(0.319)
0.001***
(0.000)
4.228***
(0.275)
5.534**
(1.152)
4.431***
(1.036)
16.951***
(1.035)
-23.680***
(2.074)

Yes
Yes
Yes
Yes
5,570
0.556

-0.050**
(0.016)
0.001**
(0.000)
0.397*
(0.180)
0.483
(0.318)
0.001***
(0.000)
4.205***
(0.273)
5.545**
(1.153)
4.388***
(1.025)
16.895***
(1.031)
-25.069***
(2.156)

Yes
Yes
Yes
Yes
5,570
0.557

𝑆_𝐵𝑂 𝐴𝑅𝐷_𝑆𝐼 𝑍 𝐸𝑠,𝑡

𝑆_𝐹𝐸 𝑀_𝐸 𝑋 𝐸𝐶𝑈𝑇 𝐼𝑉 𝐸 𝑆𝑠,𝑡

𝑆_𝑁 𝐸𝑇_𝐼 𝑁𝐶𝑂 𝑀 𝐸𝑠,𝑡

ln 𝑆_𝐸 𝑀 𝑃𝑠,𝑡

𝑆_𝐶𝑆𝑅_𝐶𝑂 𝑀𝑠,𝑡

𝑆_𝐶 𝐿𝐼_𝑅𝐼𝑆𝐾𝑆𝑠,𝑡

𝑆_𝐻𝑅_𝑃𝑂 𝐿𝐼𝐶𝑌𝑠,𝑡

Constant

Same Industry Dummies
Supplier Country FE
Supplier Industry FE
Year FE
Observations
𝑅2

0.382*
(0.180)
0.480
(0.319)
0.001***
(0.000)
4.246***
(0.274)
5.527***
(1.153)
4.459***
(1.031)
16.911***
(1.034)
-24.281***
(1.930)

Yes
Yes
Yes
Yes
5,570
0.555

0.384*
(0.180)
0.475
(0.319)
0.001***
(0.000)
4.234***
(0.274)
5.528***
(1.152)
4.447***
(1.029)
16.954***
(1.032)
-23.314***
(2.052)

Yes
Yes
Yes
Yes
5,570
0.556

0.362*
(0.179)
0.416
(0.311)
0.001***
(0.000)
4.118***
(0.273)
5.636***
(1.142)
4.454***
(1.027)
16.762***
(1.038)
-21.637***
(2.084)

Yes
Yes
Yes
Yes
5,570
0.558

Note: Note: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical significance at the 10 percent,
5 percent, 1 percent, and 0.1 percent levels. Cluster-robust standard errors (s.e.) are reported in parentheses below
parameter estimates. Multi-way clustered errors are estimated at the levels of supplier-year and buyer-year. Within 𝑅2
is reported.

82

As our baseline hypothesis, we predict a negative influence of a buyer firm’s environmental

disclosure on its supplier’s environmental disclosure. In Model 2, as presented in Table 3.5, our

empirical findings suggest a negative influence (𝛽𝐵_𝐸 𝐷 𝑏,𝑡 = -0.017, 𝑠.𝑒. = 0.012) between the

forward environmental disclosure of a supplier (𝑆_𝐸 𝐷 𝑠,𝑡+1) and the environmental disclosure of a

buying firm at time 𝑡 (𝐵_𝐸 𝐷 𝑏,𝑡); however, this impact is not statistically significant, suggesting that

𝐻1 is not supported.

In Hypothesis 𝐻2𝐴, we hypothesize that the structural equivalence between a buyer and its

supplier weakens the negative influence of the buyer on its supplier’s environmental disclosures.

Model 3, as detailed in Table 3.5, provides empirical support for this hypothesis. Our findings indi-

cate that structural equivalence between the buyer and supplier moderates the influence between the

buyer’s and supplier’s environmental disclosures in a positive direction (𝛽𝐵_𝐸 𝐷𝑏,𝑡 ×𝑆𝐷𝑠,𝑏,𝑡 = 0.722,

𝑝 < 0.01), such that the negative influence becomes weaker at higher levels of structural equiva-

lence (𝛽𝐵_𝐸 𝐷 𝑏,𝑡 = -0.033, 𝑝 < 0.05). In summary, the observed effect between the environmental

disclosure of a supplier (𝑆_𝐸 𝐷 𝑠,𝑡+1) and the environmental disclosure of a buying firm (𝐵_𝐸 𝐷 𝑏,𝑡)

supports the validity of the moderating effect of structural equivalence, as hypothesized in 𝐻2𝐴.

To provide greater specificity, our calculations suggest that a one standard deviation increase in

buyer environmental disclosure (equivalent to 19.347) is estimated to lead to a 51.379% increase

in supplier environmental disclosure (equivalent to 13.969), on average. This relationship holds

for dyads characterized by higher structural equivalence between the two entities. The percentage
increase is derived from the calculation 51.379% = 19.347×0.722

.

27.187

In Hypothesis 𝐻2𝐵, we posit that the duration of the relationship between a buyer and its supplier

weakens the negative influence between a supplier’s environmental disclosure at time 𝑡 + 1 and that

of a buying firm at time 𝑡. However, the findings presented in Model 4, as detailed in Table 3.5, do

not support this hypothesis. Our analysis reveals a non-significant and negative moderating effect

of relationship duration (𝛽𝐵_𝐸 𝐷𝑏,𝑡 ×𝑅𝐷𝑠,𝑏,𝑡 = −0.003, 𝑠.𝑒. = 0.004) on the influence between the

environmental disclosure of a supplier (𝑆_𝐸 𝐷 𝑠,𝑡+1) and the environmental disclosure of a buying

firm (𝐵_𝐸 𝐷 𝑏,𝑡), indicating that 𝐻2𝐵 lacks empirical support.

83

In Hypothesis 𝐻2𝐶, we hypothesize that the national cultural similarity between a buyer and

its supplier weakens the negative influence of the buyer on its supplier’s environmental disclo-

sures. Model 5, as detailed in Table 3.5, provides empirical evidence supporting this hypothesis.

Our results show that national cultural similarity between the buyer and supplier moderates the

influence between the buyer’s and supplier’s environmental disclosures in a positive direction

(𝛽𝐵_𝐸 𝐷 𝑏,𝑡 ×𝐶𝐷𝑠,𝑏,𝑡 = 0.001, 𝑝 < 0.01). In summary, the observed effect between the environmental

disclosure of a supplier (𝑆_𝐸 𝐷 𝑠,𝑡+1) and the environmental disclosure of a buying firm (𝐵_𝐸 𝐷 𝑏,𝑡)

provides empirical support for the moderating effect of national cultural similarity, as proposed in

Hypothesis 𝐻2𝐶. To provide a more precise analysis, we estimate that a one standard deviation

increase in buyer environmental disclosure (equivalent to 19.347) is associated with an average

increase of 0.07% in supplier disclosure (equivalent to 0.020). This effect is observed in dyads

exhibiting higher national cultural similarity between the two entities. The calculated percentage
increase is determined by 0.07% = 19.347×0.001

. In addition, we incorporate all interaction terms

27.187

in a fully specified model to jointly test all proposed hypotheses. This comprehensive approach

confirms that the results are consistent with the findings observed in Models 2 through 5.

Lastly, as shown in Figures 3.5 and 3.6, we present the interaction effects at high, middle, and

low levels of the structural and cognitive dimensions of social capital. These levels are determined

using −1 and +1 standard deviations from the mean, with the median representing the middle level.

However, for the structural dimension of social capital, since its minimum value is 0, we designate

0 as the low level in Figure 3.5.

The interaction effect, illustrated by the estimated means of supplier environmental disclosure,

suggests that the negative relationship between a supplier’s environmental disclosure (𝑆_𝐸 𝐷 𝑠,𝑡+1)

and a buyer’s environmental disclosure (𝐵_𝐸 𝐷 𝑏,𝑡) is moderated by the level of structural equiv-

alence. As shown in Figure 3.5, this moderating pattern is reflected in the differences in slopes

across varying levels of structural equivalence. Using the 𝑚𝑎𝑟𝑔𝑖𝑛𝑠 command in Stata, we examine

the slopes at low, median, and high levels of the structural dimension.

At the low level of structural equivalence (blue line), the slope is negative and statistically

84

significant (𝛽 = −0.033, 𝑝 < 0.05), indicating that as buyer environmental disclosure increases,

supplier environmental disclosure significantly decreases when structural overlap is low. At the

median level (red line), the slope remains negative and statistically significant (𝛽 = −0.024,

𝑝 < 0.1), but it is flatter compared to the low level, suggesting a weaker negative effect. At the high

level of structural equivalence (green line), the slope appears visually flat and is not statistically

significant (𝛽 = 0.002, 𝑝 > 0.1), based on margins testing. This pattern supports the existence

of a significant moderating effect, implying that greater structural equivalence between buyers and

suppliers can buffer or even neutralize the negative influence of buyer environmental disclosure on

supplier environmental disclosure.

A similar moderating pattern is observed for national cultural similarity. As depicted in Figure

3.6, the interaction effect suggests that the negative relationship between a supplier’s environmental

disclosure (𝑆_𝐸 𝐷 𝑠,𝑡+1) and a buyer’s environmental disclosure (𝐵_𝐸 𝐷 𝑏,𝑡) diminishes with increas-

ing levels of cultural similarity. For buyer–supplier pairs with low national cultural similarity (blue

line), the slope is strongly negative and statistically significant (𝛽 = −0.055, 𝑝 < 0.01), indicating

that as buyer disclosure increases, supplier disclosure significantly decreases under conditions of

greater cultural dissimilarity. At the median level of cultural similarity (red line), the slope is still

negative and statistically significant (𝛽 = −0.032, 𝑝 < 0.05), although it is flatter compared to the

low similarity group, reflecting a weakened negative effect. In contrast, at the high level of national

cultural similarity (green line), the slope is visually flat and not statistically significant (𝛽 = 0.007,

𝑝 > 0.1). This pattern further confirms the moderating role of national cultural similarity, indi-

cating that greater cultural alignment between buyers and suppliers can mitigate or eliminate the

negative impact of buyer environmental disclosure on supplier environmental disclosure.

3.6 Robustness Checks and Additional Analysis

3.6.1 Robustness Checks

In this section, drawing on previous literature that addresses potential concerns related to

endogeneity and alternative empirical specifications, we aim to validate the consistency of our

findings using the empirical setup outlined in Section 3.5.1.

85

Figure 3.5 Margins Plot of Estimated Supplier Environmental Disclosure in Model 3

Figure 3.6 Margins Plot of Estimated Supplier Environmental Disclosure in Model 5

3.6.1.1 Endogeneity

In empirical research, obtaining reliable estimates of effect magnitudes is essential for practical

relevance; however, achieving unbiased and efficient estimates is often challenged by a range of

methodological issues (Ketokivi and McIntosh, 2017). Endogeneity occurs when explanatory

variables are correlated with the error term in a regression model, which can lead to biased,

inconsistent, and inefficient parameter estimates (Ketokivi and McIntosh, 2017). In this study, we

86

take a closer look at the issue of endogeneity and discuss the strategies we employ to address it

within our research design.

First, in our empirical specification, simultaneity is unlikely to be a source of endogeneity given

our modeling approach and empirical context. To be specific, by employing a lagged structure, we

model supplier environmental disclosure as a response to buyer environmental disclosure from the

previous period, thereby reducing concerns about simultaneous determination (Diebel et al., 2024;

Gualandris et al., 2021; Dong et al., 2020). Consistent with prior literature (Diebel et al., 2024;

Villena and Dhanorkar, 2020; Jira and Toffel, 2013), our sample also reveals that buyers exhibited

a significantly higher degree of disclosure than suppliers, suggesting that buyers have historically

led by example in comparison to their suppliers.

Omitted variables pose a potential endogeneity concern, as unobserved factors may be corre-

lated with both the dependent variable and the independent variables, leading to biased estimates

(Ketokivi and McIntosh, 2017). First, our main model incorporates a range of fixed effects, includ-

ing supplier-country, supplier-industry, and year, to account for both observable and unobservable

time-invariant characteristics of suppliers, as well as potential temporal shocks. In addition, we

include a comprehensive set of control variables to further mitigate confounding influences. Nev-

ertheless, we take an additional step to address this issue by applying a two-step control function

approach.

To implement a two-step control function approach, it is essential to select instruments that

not only satisfy the relevance and exclusion conditions but are also supported by strong theoretical

justification. Prior studies have commonly used lagged explanatory variables as instruments to

address endogeneity concerns (Wang and Bellemare, 2019; Bellemare et al., 2017). In this study,

we use lagged values as instruments for the corresponding independent variables. Drawing on

prior literature (Cameron and Trivedi, 2022; Adhikary et al., 2020; Villena and Dhanorkar, 2020;

Bellemare et al., 2017; Wooldridge, 2007), our instruments satisfy both the relevance and exclusion

criteria and are grounded in strong theoretical justification. To be specific, with respect to the

relevance condition, autocorrelation in the explanatory variables suggests that the endogenous

87

variable is, to some extent, correlated with its own lag (Wang and Bellemare, 2019), and this

condition can be tested in the first-stage regressions.

On the other hand, the exclusion restriction requires that the instruments influence the dependent

variable solely through their effect on the suspected endogenous variables.

In our study, this

condition is plausibly satisfied by the use of lagged explanatory variables. To be specific, the

past environmental disclosure of downstream partners, particularly in our context, where buyers in

the manufacturing industries frequently outsource production to upstream suppliers, is unlikely to

be directly relevant to suppliers’ forward sustainability decision-making. Moreover, suppliers are

unlikely to directly incorporate the prior structural and relational characteristics of buyers’ supply

networks into their own forward-looking environmental disclosure strategies. Further, considering

the overall disclosure levels of both buyers and suppliers in our sample, much of this information

is not fully observable to suppliers at the time of decision-making. Thus, there is little reason

to believe that the lagged explanatory variables have a direct effect on the forward environmental

disclosure of suppliers. Accordingly, we use these lagged explanatory variables as instruments for

the potentially endogenous independent variables – 𝐵_𝐸 𝐷 𝑏,𝑡, 𝑆𝐷 𝑠,𝑏,𝑡, and 𝑅𝐷 𝑠,𝑏,𝑡 – in our analysis.

However, this approach cannot be applied to address the potential endogeneity of the third

social capital dimension, 𝐶𝐷 𝑠,𝑏,𝑡, as it is time-invariant in our study. Thus, we opt for a proxy for

cultural and institutional proximity. To operationalize geographical proximity at the dyadic level,

we introduce the dummy variable 𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡, which equals 1 if the buying firm and supplier

have headquarters in the same country and 0 otherwise. This metric explicitly captures whether

buyer–supplier pairs share the same national context, serving as a proxy for cultural and institutional

proximity. In this sense, it is both theoretically justified and empirically correlated with 𝐶𝐷 𝑠,𝑏,𝑡,

while the co-location of headquarters is unlikely to be theoretically related to the error term in the

supplier’s environmental disclosure equation. In addition to the lagged explanatory variables, we

employ 𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡 as an additional instrument for the final potentially endogenous independent

variable, 𝐶𝐷 𝑠,𝑏,𝑡, in our analysis. Lastly, to test the moderating effects, our primary variables of

interest are the interaction terms between buyer environmental disclosure and the three dimensions

88

of social capital. To address potential endogeneity concerns, we treat these interaction terms as the

third endogenous variable in each equation.

When we include the fitted residuals from the first-stage regressions as additional regressors

in the second-stage models – where 𝑆_𝐸 𝐷 𝑠,𝑡+1 is the dependent variable – we find no substantial

empirical differences compared to the results from our main specification with respect to hypoth-

esis testing. Specifically, using the same empirical model specification as in Table 3.5, which

incorporates control variables, fixed effects, and multi-way clustered robust standard errors, we

report the key estimates in Table 3.6. Notably, across all four models, the fitted residuals from

the first-stage reduced-form regressions are statistically insignificant in the second-stage regres-

sions, with the exception of those associated with the interaction terms involving the structural and

cognitive dimensions of social capital, both of which are marginally significant at the 10% level.

Consequently, after addressing potential endogeneity using a two-step control function approach

across all explanatory variables in our empirical models, the results suggest that endogeneity is

unlikely to pose a significant threat to the validity of our analysis.

Estimated Coefficient
𝛽𝐵_𝐸 𝐷 𝑏,𝑡
𝛽𝑆𝐷 𝑠,𝑏,𝑡
𝛽𝑅𝐷 𝑠,𝑏,𝑡
𝛽𝐶𝐷𝑠,𝑏,𝑡
𝛽𝐵_𝐸 𝐷 𝑏,𝑡 ×𝑆𝐷 𝑠,𝑏,𝑡
𝛽𝐵_𝐸 𝐷 𝑏,𝑡 ×𝑅𝐷 𝑠,𝑏,𝑡
𝛽𝐵_𝐸 𝐷 𝑏,𝑡 ×𝐶𝐷𝑠,𝑏,𝑡
Observations
𝑅2

Table 3.6 Estimates from Second-Stage Models

Model A

Model B

Model C

Model D

−0.017 (0.018) −0.035 (0.022) −0.017 (0.024) −0.054 (0.038)

–
–
–
–
–
–
2,935
0.581

−4.698 (26.280)
–
–
0.764 (0.437)+
–
–
2,935
0.583

–
0.079 (0.307)
–
–
0.000 (0.004)
–
2,935
0.581

–
–
−0.058 (0.028)*
–
–
0.001 (0.001)*
2,935
0.583

Note: +𝑝 < 0.1 and ∗𝑝 < 0.05 indicate statistical significance at the 10 percent and 5 percent levels. Cluster-robust standard errors
(s.e.) are reported in parentheses to the right of the parameter estimates. Multi-way clustered errors are estimated at the levels of
supplier-year and buyer-year. Within 𝑅2 is reported.

3.6.1.2 Multi-way Clustered Robust Standard Errors

In Section 3.5.1, we employ OLS estimation with multi-way clustered standard errors at both the

supplier-year and buyer-year levels to address potential cross-sectional correlations among dyads

89

sharing the same suppliers and buyers within each period. In this section, we refine our estimation

approach by implementing an empirical setup that utilizes different multi-way clustering strategies

to further robustify our results. Diebel et al. (2024) suggests that clustering standard errors at

the supplier’s industry-country level enables accounting for correlations arising from three nested

levels of analysis: the dyadic level (longitudinal correlations across the same dyad), the supplier

level (both longitudinal and cross-sectional correlations among dyads sharing the same supplier),

and the industry-country level (both longitudinal and cross-sectional correlations among dyads with

suppliers operating within the same industry-country pair).

As shown in Table 3.7, although there are differences in the magnitudes of the coefficients and

their cluster-robust standard errors, with respect to hypothesis testing, the findings remain consistent

with those presented in Table 3.5. Therefore, in addition to clustering at the buyer-year level, further

employing cluster-robust standard errors at the supplier’s industry-country level effectively controls

for correlations arising from three nested levels of analysis, thereby addressing more various forms

of observational correlation within our dyadic panel data.

3.6.1.3 Alternative Measure for the Cognitive Dimension of Social Capital

In Section 3.5.1, we adapt the national cultural similarity measure to capture the cognitive

dimension of social capital (𝐶𝐷 𝑠,𝑏,𝑡). Specifically, we reverse-engineer the traditional national

cultural distance metric, originally defined as the Euclidean distance between Hofstede’s cultural

dimensions of the buyer’s and supplier’s headquarters, to derive a measure of similarity rather than

distance. Thus, to test the robustness of the newly introduced metric, this section incorporates

an additional measure discussed in Section 3.6.1.1 to validate the reverse-engineered variable and

reinforce the consistency and reliability of our empirical findings.

Prior studies frequently emphasize international variations in CSR reporting, attributing these

differences primarily to national contextual factors. Variations in social and environmental re-

porting standards across countries are often driven by external pressures, stakeholder expectations,

investment demands, and long-term strategic commitments, leading to notable discrepancies in

CSR disclosure practices (Luo et al., 2013). Cultural differences across nations substantially

90

Table 3.7 Estimates of OLS Estimation with a Different Multi-way Clustering Strategy

Variable

𝐵_𝐸 𝐷 𝑏,𝑡

𝑆𝐷 𝑠,𝑏,𝑡

𝐵_𝐸 𝐷 × 𝑆𝐷

𝑅𝐷 𝑠,𝑏,𝑡

𝐵_𝐸 𝐷 × 𝑅𝐷

𝐶𝐷 𝑠,𝑏,𝑡

𝐵_𝐸 𝐷 × 𝐶𝐷

Constant

Control Variables
Same Industry Dummies
Supplier Country FE
Supplier Industry FE
Year FE
Observations
𝑅2

Model 4

Model 5

-0.010
(0.018)

0.013
(0.023)

Model 2

-0.017
(0.016)

Model 3

-0.033+
(0.019)
2.638
(27.409)
0.722+
(0.409)

0.209
(0.283)
-0.003
(0.005)

-23.314***
(3.081)

-21.637***
(3.133)

-23.680***
(3.085)

Yes
Yes
Yes
Yes
Yes
5,570
0.556

Yes
Yes
Yes
Yes
Yes
5,570
0.558

Yes
Yes
Yes
Yes
Yes
5,570
0.556

-0.050*
(0.023)
0.001*
(0.000)
-25.069***
(3.189)

Yes
Yes
Yes
Yes
Yes
5,570
0.557

Note: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical significance at the 10 percent,
5 percent, 1 percent, and 0.1 percent levels. Cluster-robust standard errors (s.e.) are reported in parentheses below
parameter estimates. Multi-way clustered errors are estimated at the supplier’s industry-country level. Within 𝑅2 is
reported

influence ethical standards, CSR orientations, organizational cultures, and managerial practices

(Pucheta-Martínez and Gallego-Álvarez, 2020; Scholtens and Dam, 2007).

On top of that, firms are more likely to collaborate with and adopt practices aligned with peers

in the same country, partly due to shared national norms, values, and institutional characteristics

(Gualandris et al., 2021; Brandon-Jones et al., 2015). In addition, non-governmental organizations

(NGOs) and other non-profit entities often monitor and publicize corporate activities concerning

environmental and social responsibility within the regions where firms operate. Within this context,

national culture and local institutional environments significantly influence communication patterns

91

between firms and stakeholders, shaping the overall framework for interactions among companies,

governments, and stakeholder groups.

Therefore, as a measure of geographical proximity at the dyadic level, we use the dummy

variable 𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡, which equals 1 if the buyer and supplier are headquartered in the same

country, and 0 otherwise, as part of this robustness check. Using the same empirical framework

outlined in Section 3.5.1, we substitute 𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡 for 𝐶𝐷 𝑠,𝑏,𝑡 and find consistent results with

the hypothesis testing presented in Section 3.5.2. To be specific, when supplier environmental

disclosure (𝑆_𝐸 𝐷 𝑠,𝑡+1) is used a dependent variable like in Section 3.5.1, our empirical results

as given as follows: 𝛽𝐵_𝐸 𝐷𝑏,𝑡 = −0.049 (𝑝 < 0.001), 𝛽𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡 = −4.563 (𝑝 < 0.001), and

𝛽𝐵_𝐸 𝐷 𝑏,𝑡 ×𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡 = 0.081 (𝑝 < 0.001).

Under the empirical framework outlined in Section 3.5.1, we further validate our findings

by introducing the alternative metric, 𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡. Specifically, we replace 𝐶𝐷 𝑠,𝑏,𝑡 with

𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡, an indicator of whether the buying firm and supplier are headquartered in the

same country, and re-estimate Model 5 in Table 3.5 using supplier environmental disclosure

(𝑆_𝐸 𝐷 𝑠,𝑡+1) as the dependent variable. Despite differences in the magnitude of the coefficients,

our empirical results in Section 3.5.2 remain consistent with those derived from 𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡,

confirming robust and statistically significant relationships: 𝛽𝐵_𝐸 𝐷 𝑏,𝑡 = −0.049 (𝑝 < 0.001),

𝛽𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡 = −4.563 (𝑝 < 0.001), and 𝛽𝐵_𝐸 𝐷 𝑏,𝑡 ×𝑆 𝐴𝑀 𝐸_𝐻𝑄𝑠,𝑏,𝑡 = 0.081 (𝑝 < 0.001). Put

differently, cultural and institutional proximity between the buyer’s and supplier’s headquarters

moderates the negative impact between their environmental disclosures in a positive direction,

with the negative influence diminishing as the cultural and institutional proximity increases. In

summary, although the magnitudes of the coefficients and their cluster-robust standard errors vary,

the empirical findings for hypothesis testing remain consistent with those presented in Table 3.5.

3.6.2 Additional Analysis

In this section, we compare the levels of supplier environmental disclosure across two buyer

industries – Materials and Pharmaceuticals – given that manufacturers in both sectors are particu-

larly sensitive to environmental disclosures due to their focus on safety, environmental impact, and

92

sustainability. Our goal is to determine whether there are differences in environmental disclosure

among suppliers serving buyers across these distinct industry sectors.

3.6.2.1

Industry Comparison

When an analysis of variance (ANOVA) yields a significant result, it indicates that at least

one group statistically differs from the others. To further explore these differences, ANOVA

is typically followed by post hoc tests that focus on specific comparisons, commonly involving

pairwise comparisons of means, which help to identify which specific groups differ from each

other (Abdi and Williams, 2010).

When employing a one-way fixed-effects ANOVA to model data, we operate under four fun-

damental assumptions: 1.

independence of observations, 2. additivity of effects, 3. normality

of residuals, and 4. homogeneity of variances (Larson, 2008). Specifically, the fourth assump-

tion mandates that the within-group random errors exhibit identical variance across all treatment

groups. To verify this, Levene’s test and Bartlett’s test are commonly employed, with the choice

of test depending on the normality of the data (Larson, 2008). Thus, we evaluate the homogeneity

of variances using Levene’s test and Bartlett’s test, and determine that the assumption of equal

variances across the buyers’ industries is not met.

Given the results of Levene’s test and Bartlett’s test, we try the Kruskal-Wallis test (Kruskal and

Wallis, 1952), Mann–Whitney test (Mann and Whitney, 1947), and Dunn’s pairwise comparison

test (Dunn, 1964) to check for any difference between supplier environmental disclosures of these

two groups. The Kruskal-Wallis test is a non-parametric alternative to one-way ANOVA, extending

the Mann-Whitney U test to compare differences among three or more independent groups on

a continuous variable that does not follow a normal distribution (McKight and Najab, 2010).

Given that the Kruskal-Wallis test is an omnibus test for median differences, rejecting the null

hypothesis typically warrants conducting multiple pairwise comparisons to further examine the

specific differences in medians (Dinno, 2015).

As shown in Table 3.8, the Kruskal-Wallis test provides evidence that rejects the null hypothesis,

indicating that the samples do not come from the same population (𝜒2 adjusted for ties = 496.882,

93

Table 3.8 Kruskal-Wallis and Wilcoxon Rank-Sum Test Results

Buyer Industry

Observations Rank Sum

Materials Industry
Pharmaceutical Industry
Combined

6,331
2,965
9,296

32,107,891
11,104,566
43,212,456

𝑝 < 0.001) (Kruskal and Wallis, 1952), while the Mann-Whitney rank-sum test rejects the null

hypothesis (𝑍 − 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 = 22.291, 𝑝 < 0.001) that two independent (unmatched) samples are

drawn from populations with identical distributions (Mann and Whitney, 1947). In addition, Dunn’s

test of multiple comparisons using mean ranks also indicates a significant difference between the

buyers’ industries in terms of their supplier environmental disclosure, and this difference is highly

significant (𝑍 − 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 = 22.291, 𝑝 < 0.001). In summary, based on the results in Table 3.8, the

mean rank for the materials industry is significantly higher than that of the pharmaceutical industry,

suggesting that the materials industry likely exhibits a larger supplier environmental disclosure.

3.7 Discussion and Conclusion

This study underscores the importance of social capital dimensions embedded within BSRs in

shaping the level of supplier environmental disclosure, particularly in the context of supply chain

leakage. In the manufacturing industry, not all suppliers within a network are equally willing to

disclose their environmental performance. Such reluctance may arise from a range of factors, which

we interpret through the lens of the PHH. Consequently, and somewhat counterintuitively, higher

levels of buyer environmental disclosure may be associated with lower levels of supplier disclosure.

In addition, various firm-level and supply chain-level characteristics that shape visibility across

supply chains may influence the negative impact between a buyer’s and a supplier’s environmen-

tal disclosures. Building on prior literature grounded in social capital theory, we examine the

moderating effects of three key dimensions of social capital, structural equivalence, relationship

duration, and cultural similarity, which capture essential relational attributes embedded within

buyer–supplier dyads in the materials and pharmaceutical industries.

Drawing on voluntary environmental disclosure data from the Bloomberg ESG database, we

94

first investigate the influence between a supplier’s environmental disclosure and that of its buying

firm across two manufacturing sectors. Our analysis reveals that, although the overall relationship

is negative but not statistically significant, it becomes both negative and significant under conditions

of low and median structural equivalence. In contrast, when structural equivalence is high, the

negative effect is nullified and becomes non-significant, suggesting that the influence of a buyer’s

environmental disclosure on that of its suppliers is more likely to be voluntary when supply

chain visibility is greater. However, we do not find evidence supporting a moderating effect

of relationship duration. Lastly, we found that cultural and institutional proximity between the

buyer’s and supplier’s headquarters moderates the negative influence of a buyer on its supplier’s

environmental disclosures in a positive direction, with the negative influence diminishing as the

cultural and institutional proximity increases.

Our findings highlight the critical role of BSRs in shaping environmental disclosure within

SSCM. For buyers seeking to promote comprehensive upstream transparency, rather than inad-

vertently shifting and obscuring environmental harm, it is essential to invest in building structural

capital and shared cognitive frameworks with suppliers. In the absence of such alignment, a buyer’s

own environmental disclosure may paradoxically lead to reduced disclosure by suppliers. By ex-

amining this dynamic in the materials and pharmaceutical industries – both subject to heightened

environmental scrutiny – we demonstrate that buyer-led transparency can, under certain conditions,

contribute to hidden pollution upstream. However, when buyers and suppliers are well-aligned and

supply chain visibility is high, this negative effect can be mitigated or even eliminated.

3.7.1 Theoretical Implications

To meet the growing and diverse demands of stakeholders, firms nowadays collaborate with

members in supply networks to leverage partners’ competency in key specialized areas. In supply

networks, most of the buying firms tend to have many suppliers, enabling them to outsource some of

their business functions, such as production, services, and finances, to external upstream partners.

In many cases, buying firms engage proactively in these relationships to generate joint value, such

as enhanced productivity and innovation, that can ultimately improve their operational performance

95

(Kim and Choi, 2018; Krause et al., 2007). At the same time, participants in supply networks are

increasingly pressured to comply with the focal firm’s initiatives related to production processes,

material handling, and product distribution. From an environmental perspective, firms in the

manufacturing industry often respond to these competing demands by strategically locating their

production facilities in countries with more lenient environmental regulations, thereby minimizing

pollution abatement costs (Berry et al., 2021). This contradicts the prevailing belief that increased

environmental disclosure by a buying firm serves as a catalyst for similar behavior among its

suppliers.

Our primary theoretical contributions examine whether the logic of the supply chain leakage

effect, often associated with the PHH, can be extended to corporate environmental disclosures of

buyers and suppliers. This is because supply chain leakage may be linked to a firm’s efforts to op-

timize operations and manage environmental externalities through the outsourcing of operational

activities, ultimately affecting the environmental performance of the entire supply chain (Song

et al., 2023). Building on the underlying logic of supply chain leakage and the PHH, we propose a

negative influence between a supplier’s environmental disclosure and that of its buying firm in the

manufacturing industry. This may occur when a focal firm reduces its own carbon emissions while

promoting a pristine environmental image through disclosure, potentially masking the relocation or

outsourcing of polluting activities to upstream suppliers. Such carbon leakage can impose a dispro-

portionate environmental burden on suppliers, resulting in higher upstream carbon footprints. Our

findings indicate that while the overall influence between a buyer’s and a supplier’s environmental

disclosures is negative but statistically non-significant, it becomes both negative and significant

under conditions of low and median structural equivalence. Although Song et al. (2023) provide

evidence of supply chain leakage by documenting a negative relationship between a firm’s internal

GHG emissions intensity and that of its suppliers, empirically validating the direct application

of the PHH remains particularly challenging in the context of voluntary corporate environmental

disclosure (Berry et al., 2021).

Our additional theoretical contributions investigate how diverse social capital factors embedded

96

within supply networks shape supplier environmental disclosure behavior. Despite the challenges

posed by supply chain leakage, supply chains themselves can serve as valuable environmental

assets by contributing to the mitigation of environmental issues across the broader network (Song

et al., 2023). A substantial body of literature in SCM highlights that BSRs within supply networks,

comprising numerous actors, can take various forms depending on how and why the relationships

are established. From a strategic standpoint, firms often prefer to cultivate long-term, cooperative

relationships with key suppliers, as purchasing is recognized as a critical lever for gaining competi-

tive advantage (Carr and Pearson, 1999). However, BSRs can range from adversarial to cooperative

in nature (Kim and Choi, 2015; Carr and Pearson, 1999; Carter et al., 1998). Adversarial BSRs are

typically characterized by multiple sourcing strategies and shorter-term contracts (Carr and Pear-

son, 1999; Carter et al., 1998), whereas cooperative BSRs involve longer-term, closely integrated

partnerships aimed at mutual benefit (Kim and Choi, 2015; Carter et al., 1998). Based on our

analysis, suppliers engaged in more arm’s length relationships characterized by lower social capital

may be more reluctant to disclose, potentially to avoid reputational or commercial risks. In such

cases, where suppliers have fewer overlapping partners and lower cultural similarity with the buyer,

supply chain visibility tends to be weaker.

In today’s hypercompetitive business environment and amid the evolving dynamics of BSRs,

firms are increasingly adopting more nuanced and sophisticated relational strategies (Kim and

Choi, 2015). Drawing on social capital theory, alongside the concepts of supply chain leakage and

the PHH, this study extends prior research by examining how various dimensions of social capital

influence the effect of buyer behavior on supplier disclosure. Our work complements existing

studies on network effects in supply chains (Chae et al., 2022; Adhikary et al., 2020; Bellamy

et al., 2020; Sharma et al., 2020) by showing that social capital embedded within BSRs shapes the

influence between a buyer’s and a supplier’s environmental disclosures. In doing so, our findings

help explain why suppliers may not necessarily mirror a buyer’s public environmental commitments.

Instead, strong buyer disclosure can, under certain conditions, lead to lower supplier disclosure

– a manifestation of the supply chain leakage effect – depending on the structural and cognitive

97

dimensions of social capital. Strengthening these dimensions within BSRs is thus essential for

achieving authentic environmental progress across supply networks, rather than merely shifting

emissions and obscuring upstream impacts.

Given the prevalence of firm-level studies, recent literature has increasingly called for more

research focused on dyadic-level analysis in SSCM. Drawing on detailed data on buyer and supplier

environmental disclosure, particularly within manufacturing supply networks, this study contributes

to the literature by providing a granular understanding of the relational mechanisms through

which buyers influence supplier behavior in the context of environmental disclosure.

In doing

so, it advances theoretical discussions on BSRs by integrating social capital theory with emerging

concerns around sustainability and supply chain governance.

3.7.2 Managerial Implications

Our findings provide practical insights into corporate voluntary environmental disclosure from

a network-based perspective. While prior literature often suggests that increased environmental

disclosure by a buying firm encourages suppliers to follow suit, publicly signaling green objec-

tives and fostering collective sustainability efforts across the supply chain, this assumption is not

universally supported. Some studies argue that firms may strategically employ environmental dis-

closure as a form of greenwashing, shifting their environmental burdens upstream to suppliers in

response to performance and disclosure pressures (Song et al., 2023; Dai et al., 2021; Kim and

Lyon, 2011). Accordingly, and somewhat paradoxically, increased buyer environmental disclosure

may correspond with reduced disclosure by suppliers.

Based on the PHH, buyers, particularly in the manufacturing industry, may outsource their

operational and production activities, which can negatively impact suppliers’ environmental per-

formance. As a result, suppliers may be less inclined to publicly disclose their environmental

performance. Therefore, given that strong environmental performers tend to disclose more infor-

mation than their weaker counterparts (Al-Tuwaijri et al., 2004), we recommend that buyers in the

manufacturing industry adopt a moral foundation to promote environmental transparency across

their supply networks (Diebel et al., 2024). Buying firms often serve as catalysts for environmental

98

responsibility, encouraging their suppliers to adopt sustainable practices in alignment with the

firms’ own expectations and demands (Jira and Toffel, 2013; Lee and Klassen, 2008).

We advise that buyers recognize the significant role their supply network structure and partner

characteristics play in shaping supplier environmental disclosure. The configuration of a buyer’s

network relationships critically influences suppliers’ motivation to engage in sustainability initia-

tives. Specifically, suppliers with a higher number of overlapping partners with the buyer are more

likely to align with the buyer’s environmental practices and participate in sustainability efforts.

Moreover, collaboration with culturally similar suppliers tends to be more effective, as greater trust

and shared understanding foster stronger relationships and enhance the likelihood of compliance

with disclosure expectations. In such culturally aligned partnerships, buyers can anticipate greater

willingness from suppliers to publicly disclose environmental performance, thereby strengthening

collaborative efforts to address negative environmental externalities and improving overall supply

chain visibility. However, in today’s dynamic and competitive business landscape, buyers often

pursue innovation by engaging with less familiar and culturally divergent suppliers (Chae et al.,

2020). In these cases, buyers should be aware that such suppliers may be less inclined to conform

to environmental disclosure expectations. To improve environmental outcomes, it is essential for

buyers to invest in building trust and collaboration with these suppliers, enabling more effective

management of upstream environmental risks.

Finally, as stakeholders increasingly demand that firms and their supply networks account

for and publicly disclose their environmental impacts and associated risks, it becomes crucial for

policymakers to understand the drivers that motivate suppliers to engage in environmental disclosure

(Diebel et al., 2024). In general, firms are more likely to collaborate and adopt practices that align

with those of peers who share similar values and characteristics (Gualandris et al., 2021; Brandon-

Jones et al., 2015). This tendency is partly driven by the presence of shared traits among firms

operating within similar industry or institutional contexts. Further, non-governmental organizations

(NGOs) and other non-profit entities often monitor and publicize corporate environmental and social

responsibility activities in the regions where companies operate. However, organizations focused

99

on monitoring supply chain-related environmental and social practices often face limitations, as

their efforts are constrained by local contexts, norms, and regulatory environments (Sodhi and Tang,

2019). Within this landscape, national culture and institutional environments play a pivotal role in

shaping how firms communicate with stakeholders and in structuring the interactions among firms,

governments, and civil society. Therefore, we encourage policymakers to consider the distinct

characteristics of firms across various industry sectors and supply network configurations when

formulating sustainability policies, as these firm-level and supply chain-level attributes, embedded

within BSRs, are essential for fostering and sustaining meaningful environmental progress across

the entire supply network.

3.7.3 Limitations and Future Research Directions

Although this study offers a detailed empirical research design and employs various precautions,

such as robustness checks and supplementary analyses, it still has several limitations. First, since

our analysis is limited to buyers within the materials and pharmaceutical industries, the findings

may offer limited insights for other industry contexts. In other words, the empirical results may not

be fully generalizable to buyers operating outside the manufacturing sector.

As highlighted by Chae et al. (2020), this study employs a relatively simple measure of the

cognitive dimension of social capital. Specifically, we transform national cultural distance into

national cultural similarity by multiplying it by negative one. This transformation allows for a more

context-sensitive interpretation, enhancing the measure’s ability to capture the nuances of cognitive

alignment, or misalignment, in international business relationships. Nevertheless, the cognitive

dimension of social capital remains a relatively simple and time-invariant construct within each

BSR.

Third, our empirical analyses primarily rely on the FactSet Revere supply chain relationships

dataset, complemented by the Bloomberg ESG dataset. As noted in Section 3.4.1, the FactSet

Revere dataset covers over 31,000 publicly traded companies worldwide and includes more than

450,000 documented business relationships (FactSet Revere, 2021). However, regardless of the

data source, mapping a firm’s supply network is an ongoing process. As such, we acknowledge

100

that certain supply chain relationships may be missing from our dataset or may have evolved over

time (Song et al., 2023; Bellamy et al., 2020).

This study also opens several promising avenues for future research. In relation to the first and

third limitations, the generalizability of our findings is constrained by the exclusive focus on two

manufacturing industries and the incomplete nature of the network data. Future research could

enhance external validity by expanding the industry scope and utilizing more comprehensive and

dynamic supply network data.

Supply chains can fundamentally serve as strategic environmental assets by facilitating the

mitigation of environmental issues across broader business networks (Song et al., 2023). Building

on this perspective, future research could explore a wider range of firm performance dimensions

within supply networks or dyadic relationships, extending beyond CSR activities to encompass

innovation, economic outcomes, and operational efficiency (Bellamy et al., 2020; Chae et al., 2020;

Sharma et al., 2020; Lu and Shang, 2017). In addition, although this study primarily focuses on

manufacturing industries, where the supply chain leakage effect serves as a key mechanism, firms

across sectors have increasingly advanced both internal and external environmental performance

through collaboration with suppliers. Therefore, it would be valuable for future research to examine

companies that emphasize integration, strategic partnerships, and collaborative practices as core

mechanisms for fostering mutual benefits and achieving shared sustainability objectives between

buyers and suppliers.

Conventional wisdom has long held that increased environmental disclosure by a buying firm

should encourage its suppliers to follow suit. However, this study challenges that assumption by

illustrating how suppliers may not mirror a buyer’s public environmental commitments, particularly

in the presence of the supply chain leakage effect, where strong buyer disclosure can, paradoxically,

lead to lower supplier disclosure. In addition, by leveraging the reciprocal nature of BSRs, this

research contributes to the relatively underexplored domain of social capital within BSRs.

In

summary, we extend the application of social capital theory to the context of SSCM, demonstrating

that strengthening key social capital dimensions is essential for fostering authentic environmental

101

progress across the supply network, rather than simply shifting emissions or obscuring upstream

impacts.

102

REFERENCES

Abdi, H. and Williams, L. J. (2010). Tukey’s honestly significant difference (HSD) test. Encyclo-

pedia of Research Design, 3(1):1–5.

Adhikary, A., Sharma, A., Diatha, K. S., and Jayaram, J. (2020). Impact of buyer-supplier network
complexity on firms’ greenhouse gas (ghg) emissions: An empirical investigation. International
Journal of Production Economics, 230:107864.

Al-Tuwaijri, S. A., Christensen, T. E., and Hughes Ii, K. (2004). The relations among environmental
disclosure, environmental performance, and economic performance: A simultaneous equations
approach. Accounting, Organizations and Society, 29(5-6):447–471.

Andersson, U., Forsgren, M., and Holm, U. (2002). The strategic impact of external networks:
Subsidiary performance and competence development in the multinational corporation. Strategic
Management Journal, 23(11):979–996.

Autry, C. W. and Golicic, S. L. (2010). Evaluating buyer–supplier relationship–performance spirals:

A longitudinal study. Journal of Operations Management, 28(2):87–100.

Autry, C. W. and Griffis, S. E. (2008). Supply chain capital: The impact of structural and relational

linkages on firm execution and innovation. Journal of Business Logistics, 29(1):157–173.

Awaysheh, A. and Klassen, R. D. (2010). The impact of supply chain structure on the use of supplier
socially responsible practices. International Journal of Operations & Production Management,
30(12):1246–1268.

Basole, R. C. and Bellamy, M. A. (2014). Supply network structure, visibility, and risk diffusion:

A computational approach. Decision Sciences, 45(4):753–789.

Basole, R. C., Ghosh, S., and Hora, M. S. (2017). Supply network structure and firm perfor-
mance: Evidence from the electronics industry. IEEE Transactions on Engineering Management,
65(1):141–154.

Bateman, A. and Bonanni, L. (2019). What supply chain transparency really means. Harvard

Business Review, 20:2–8.

Bellamy, M. A., Dhanorkar, S., and Subramanian, R. (2020). Administrative environmental in-
novations, supply network structure, and environmental disclosure. Journal of Operations
Management, 66(7-8):895–932.

Bellemare, M. F., Masaki, T., and Pepinsky, T. B. (2017). Lagged explanatory variables and the

estimation of causal effect. The Journal of Politics, 79(3):949–963.

Berry, H., Kaul, A., and Lee, N. (2021). Follow the smoke: The pollution haven effect on global

103

sourcing. Strategic Management Journal, 42(13):2420–2450.

Beugelsdijk, S., Kostova, T., Kunst, V. E., Spadafora, E., and Van Essen, M. (2018). Cultural
distance and firm internationalization: A meta-analytical review and theoretical implications.
Journal of Management, 44(1):89–130.

Bloomberg Professional Services (2018). Environmental, social & governance (ESG) product:
Comprehensive & accurate analysis on material financial risks & opportunities. https://data.
bloomberglp.com/professional/sites/10/1148330431.pdf.

Blount, I. and Li, M. (2021). How buyers’ attitudes toward supplier diversity affect their expenditures

with ethnic minority businesses. Journal of Supply Chain Management, 57(3):3–24.

Bolino, M. C., Turnley, W. H., and Bloodgood, J. M. (2002). Citizenship behavior and the creation

of social capital in organizations. Academy of Management Review, 27(4):505–522.

Bourlakis, M., Maglaras, G., Aktas, E., Gallear, D., and Fotopoulos, C. (2014). Firm size and
sustainable performance in food supply chains: Insights from greek SMEs. International Journal
of Production Economics, 152:112–130.

Brammer, S. and Pavelin, S. (2008). Factors influencing the quality of corporate environmental

disclosure. Business Strategy and the Environment, 17(2):120–136.

Brandon-Jones, E., Squire, B., and Van Rossenberg, Y. G. (2015). The impact of supply base
complexity on disruptions and performance: The moderating effects of slack and visibility.
International Journal of Production Research, 53(22):6903–6918.

Burt, R. S. (1987). Social contagion and innovation: Cohesion versus structural equivalence.

American Journal of Sociology, 92(6):1287–1335.

Burt, R. S. (1992). Structural holes. Harvard University Press.

Burt, R. S. (1997). The contingent value of social capital. Administrative Science Quarterly, pages

339–365.

Burt, R. S. (2004). Structural holes and good ideas. American Journal of Sociology, 110(2):349–

399.

Cameron, A. C., Gelbach, J. B., and Miller, D. L. (2011). Robust inference with multiway clustering.

Journal of Business & Economic Statistics, 29(2):238–249.

Cameron, A. C. and Trivedi, P. K. (2022). Microeconometrics using STATA, volumes I and II.

Campbell, J. L. (2007). Why would corporations behave in socially responsible ways? an institu-
tional theory of corporate social responsibility. Academy of Management Review, 32(3):946–967.

104

Cannizzaro, A. P. (2020). Social influence and MNE strategic response to political risk: A global

network approach. Journal of International Business Studies, 51(5):829–850.

Capaldo, A. (2007). Network structure and innovation: The leveraging of a dual network as a

distinctive relational capability. Strategic Management Journal, 28(6):585–608.

Carr, A. S. and Pearson, J. N. (1999). Strategically managed buyer–supplier relationships and

performance outcomes. Journal of Operations Management, 17(5):497–519.

Carter, J. R., Smeltzer, L., and Narasimhan, R. (1998). The role of buyer and supplier relationships
in integrating TQM through the supply chain. European Journal of Purchasing & Supply
Management, 4(4):223–234.

Chae, S., Son, B.-G., Yan, T., and Yang, Y. S. (2022). Supply chains and the success of M&As:
Investigating the effect of structural equivalence of merging firms’ supplier and customer bases.
International Journal of Operations & Production Management, 42(8):1272–1293.

Chae, S., Yan, T., and Yang, Y. (2020). Supplier innovation value from a buyer–supplier structural
equivalence view: Evidence from the PACE awards in the automotive industry. Journal of
Operations Management, 66(7-8):820–838.

Chedid, F., Kocabasoglu-Hillmer, C., and Ries, J. M. (2021). The interaction between supply
networks and internal networks: Performance implications. International Journal of Operations
& Production Management, 41(6):860–881.

Chen, C., Jiang, D., and Li, W. (2023). Keeping up with the CSR joneses: The impact of industry
peers on focal firms’ CSR performance. Humanities and Social Sciences Communications,
10(1):1–12.

Cheung, Y.-L., Haw, I.-M., Hu, B., Swink, M., and Zhang, W. (2020). Common institutional
investors and supplier performance in supply chains. Journal of Operations Management,
66(6):670–696.

Cho, C. H., Guidry, R. P., Hageman, A. M., and Patten, D. M. (2012). Do actions speak louder
than words? An empirical investigation of corporate environmental reputation. Accounting,
Organizations and Society, 37(1):14–25.

Cho, C. H. and Patten, D. M. (2007). The role of environmental disclosures as tools of legitimacy:

A research note. Accounting, Organizations and Society, 32(7-8):639–647.

Christensen, H. B., Hail, L., and Leuz, C. (2021). Mandatory CSR and sustainability reporting:

Economic analysis and literature review. Review of Accounting Studies, 26(3):1176–1248.

Claridge, T. (2018). Dimensions of social capital-structural, cognitive, and relational. Social

Capital Research, 1:1–4.

105

Clarkson, P. M., Li, Y., Richardson, G. D., and Vasvari, F. P. (2008). Revisiting the relation between
environmental performance and environmental disclosure: An empirical analysis. Accounting,
Organizations and Society, 33(4-5):303–327.

Coleman, J. S. (1994). Foundations of social theory. Harvard University Press.

Cormier, D. and Magnan, M. (1999). Corporate environmental disclosure strategies: Determinants,

costs and benefits. Journal of Accounting, Auditing & Finance, 14(4):429–451.

Cowen, S. S., Ferreri, L. B., and Parker, L. D. (1987). The impact of corporate characteristics
on social responsibility disclosure: A typology and frequency-based analysis. Accounting,
Organizations and Society, 12(2):111–122.

Cox, A., Sanderson, J., and Watson, G. (2001). Supply chains and power regimes: Toward
an analytic framework for managing extended networks of buyer and supplier relationships.
Journal of Supply Chain Management, 37(2):28.

Culot, G., Podrecca, M., Nassimbeni, G., Orzes, G., and Sartor, M. (2023). Using supply chain
databases in academic research: A methodological critique. Journal of Supply Chain Manage-
ment, 59(1):3–25.

Dai, R., Liang, H., and Ng, L. (2021). Socially responsible corporate customers. Journal of

Financial Economics, 142(2):598–626.

D’Amico, E., Coluccia, D., Fontana, S., and Solimene, S. (2016). Factors influencing corporate

environmental disclosure. Business Strategy and the Environment, 25(3):178–192.

Delgado-Márquez, B. L., Pedauga, L. E., and Cordón-Pozo, E. (2017). Industries regulation and
firm environmental disclosure: A stakeholders’ perspective on the importance of legitimation
and international activities. Organization & Environment, 30(2):103–121.

Diebel, W., Gualandris, J., and Klassen, R. D. (2024). How do suppliers respond to institutional
complexity? Examining voluntary public environmental disclosure in a global manufacturing
supply network. Journal of Operations Management.

Dinno, A. (2015). Nonparametric pairwise multiple comparisons in independent groups using

Dunn’s test. The Stata Journal, 15(1):292–300.

Dong, Y., Skowronski, K., Song, S., Venkataraman, S., and Zou, F. (2020). Supply base innovation

and firm financial performance. Journal of Operations Management, 66(7-8):768–796.

Dunn, O. J. (1964). Multiple comparisons using rank sums. Technometrics, 6(3):241–252.

Ebbes, P. and Netzer, O. (2021). Using social network activity data to identify and target job

seekers. Management Science.

106

Elkington, J. and Rowlands, I. H. (1999). Cannibals with forks: The triple bottom line of 21st

century business. Alternatives Journal, 25(4):42.

Eskeland, G. S. and Harrison, A. E. (2003). Moving to greener pastures? Multinationals and the

pollution haven hypothesis. Journal of Development Economics, 70(1):1–23.

FactSet Revere (2021). Supply chain relationships data and methodology guide. Technical report,

FactSet Research Systems Inc.

Fallan, E. (2016). Environmental reporting regulations and reporting practices. Social and Envi-

ronmental Accountability Journal, 36(1):34–55.

Fleming, L., King III, C., and Juda, A. I. (2007). Small worlds and regional innovation. Organization

Science, 18(6):938–954.

Friedkin, N. E. (1998). A structural theory of social influence. Cambridge University Press.

González, P., Sarkis, J., and Adenso-Díaz, B. (2008). Environmental management system cer-
tification and its influence on corporate practices: Evidence from the automotive industry.
International Journal of Operations & Production Management, 28(11):1021–1041.

Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness.

American Journal of Sociology, 91(3):481–510.

Granovetter, M. S. (1973). The strength of weak ties. American Journal of Sociology, 78(6):1360–

1380.

Grimm, J. H., Hofstetter, J. S., and Sarkis, J. (2014). Critical factors for sub-supplier management:
A sustainable food supply chains perspective. International Journal of Production Economics,
152:159–173.

Gualandris, J., Klassen, R. D., Vachon, S., and Kalchschmidt, M. (2015). Sustainable evaluation and
verification in supply chains: Aligning and leveraging accountability to stakeholders. Journal of
Operations Management, 38:1–13.

Gualandris, J., Longoni, A., Luzzini, D., and Pagell, M. (2021). The association between sup-
ply chain structure and transparency: A large-scale empirical study. Journal of Operations
Management, 67(7):803–827.

Haniffa, R. M. and Cooke, T. E. (2005). The impact of culture and governance on corporate social

reporting. Journal of Accounting and Public Policy, 24(5):391–430.

Hansen, M. T. (1999). The search-transfer problem: The role of weak ties in sharing knowledge

across organization subunits. Administrative Science Quarterly, 44(1):82–111.

107

Hardcopf, R., Shah, R., and Dhanorkar, S. (2021). The impact of a spill or pollution accident on firm
environmental activity: An empirical investigation. Production and Operations Management,
30(8):2467–2491.

Hardcopf, R., Shah, R., and Mukherjee, U. (2019). Explaining heterogeneity in environmental man-
agement practice adoption across firms. Production and Operations Management, 28(11):2898–
2918.

Hassan, A. (2023). Social and environmental accounting research in vulnerable and exploitable

less-developed countries: A theoretical extension. Accounting Forum, 47(3):365–389.

Hoetker, G. (2005). How much you know versus how well I know you: Selecting a supplier for a

technically innovative component. Strategic Management Journal, 26(1):75–96.

Hoetker, G., Swaminathan, A., and Mitchell, W. (2007). Modularity and the impact of buyer–

supplier relationships on the survival of suppliers. Management Science, 53(2):178–191.

Hofer, C., Cantor, D. E., and Dai, J. (2012). The competitive determinants of a firm’s environmental
management activities: Evidence from us manufacturing industries. Journal of Operations
Management, 30(1-2):69–84.

Hofstede, G. (2011). Dimensionalizing cultures: The hofstede model in context. Online Readings

in Psychology and Culture, 2(1):8.

Hu, Y. (2005). Efficient, high-quality force-directed graph drawing. Mathematica journal,

10(1):37–71.

Huang, S., Du, P., Hong, Y., and Wu, W. (2023). Are corporate environmental, social, and
governance practices contagious? The peer-effect perspective. Energy & Environment, page
0958305X231177749.

Hummel, K. and Schlick, C. (2016). The relationship between sustainability performance and
sustainability disclosure–reconciling voluntary disclosure theory and legitimacy theory. Journal
of Accounting and Public Policy, 35(5):455–476.

Ireland, R. D. and Webb, J. W. (2007). A multi-theoretic perspective on trust and power in strategic

supply chains. Journal of Operations Management, 25(2):482–497.

Jira, C. and Toffel, M. W. (2013). Engaging supply chains in climate change. Manufacturing &

Service Operations Management, 15(4):559–577.

Ketokivi, M. and McIntosh, C. N. (2017). Addressing the endogeneity dilemma in operations man-
agement research: Theoretical, empirical, and pragmatic considerations. Journal of Operations
Management, 52:1–14.

108

Kim, D. Y. (2014). Understanding supplier structural embeddedness: A social network perspective.

Journal of Operations Management, 32(5):219–231.

Kim, E.-H. and Lyon, T. P. (2011). Strategic environmental disclosure: Evidence from the DOE’s
voluntary greenhouse gas registry. Journal of Environmental Economics and Management,
61(3):311–326.

Kim, Y. and Choi, T. Y. (2015). Deep, sticky, transient, and gracious: An expanded buyer–supplier

relationship typology. Journal of Supply Chain Management, 51(3):61–86.

Kim, Y. and Choi, T. Y. (2018). Tie strength and value creation in the buyer-supplier context: A
U-Shaped relation moderated by dependence asymmetry. Journal of Management, 44(3):1029–
1064.

Kim, Y., Choi, T. Y., Yan, T., and Dooley, K. (2011). Structural investigation of supply networks:

A social network analysis approach. Journal of Operations Management, 29(3):194–211.

Kim, Y. H. and Henderson, D. (2015). Financial benefits and risks of dependency in triadic supply

chain relationships. Journal of Operations Management, 36:115–129.

Koberg, E. and Longoni, A. (2019). A systematic review of sustainable supply chain management

in global supply chains. Journal of Cleaner Production, 207:1084–1098.

Kogut, B. and Singh, H. (1988). The effect of national culture on the choice of entry mode. Journal

of International Business Studies, 19(3):411–432.

Kossinets, G. and Watts, D. J. (2006). Empirical analysis of an evolving social network. Science,

311(5757):88–90.

Kraft, T., Valdés, L., and Zheng, Y. (2018). Supply chain visibility and social responsibility: Inves-
tigating consumers’ behaviors and motives. Manufacturing & Service Operations Management,
20(4):617–636.

Kraft, T. and Zheng, Y. (2021). How supply chain transparency boosts business value. MIT Sloan

Management Review, 63(1):34–40. Reprint #63105.

Krause, D. R., Handfield, R. B., and Tyler, B. B. (2007). The relationships between supplier
development, commitment, social capital accumulation and performance improvement. Journal
of Operations Management, 25(2):528–545.

Kruskal, W. H. and Wallis, W. A. (1952). Use of ranks in one-criterion variance analysis. Journal

of the American Statistical Association, 47(260):583–621.

Kumar, S., Narayanan, S., and Salvador, F. (2020). Innovation in supply networks—a research

framework and roadmap. Journal of Operations Management, 66(7-8):754–767.

109

Larson, M. G. (2008). Analysis of variance. Circulation, 117(1):115–121.

Leary, M. T. and Roberts, M. R. (2014). Do peer firms affect corporate financial policy? The

Journal of Finance, 69(1):139–178.

Lee, C.-H. and Ha, B.-C. (2018). The impact of buyer-supplier relationships’ social capital on bi-
directional information sharing in the supply chain. Journal of Business & Industrial Marketing,
33(3):325–336.

Lee, H. L. (2010). Don’t tweak your supply chain–rethink it end to end. Harvard Business Review,

88(10):62–69.

Lee, S.-Y. and Klassen, R. D. (2008). Drivers and enablers that foster environmental management
capabilities in small-and medium-sized suppliers in supply chains. Production and Operations
Management, 17(6):573–586.

Levinson, A. and Taylor, M. S. (2008). Unmasking the pollution haven effect.

International

Economic Review, 49(1):223–254.

Lin, N. (2017). Building a network theory of social capital. Social Capital, pages 3–28.

Lin, Y., Mao, Y., and Wang, Z. (2018).

Institutional ownership, peer pressure, and voluntary

disclosures. The Accounting Review, 93(4):283–308.

Linton, J. D., Klassen, R., and Jayaraman, V. (2007). Sustainable supply chains: An introduction.

Journal of Operations Management, 25(6):1075–1082.

Liu, C. (2018). Are women greener? Corporate gender diversity and environmental violations.

Journal of Corporate Finance, 52:118–142.

Locke, R. M., Qin, F., and Brause, A. (2007). Does monitoring improve labor standards? Lessons

from Nike. Ilr Review, 61(1):3–31.

Lu, G. and Shang, G. (2017). Impact of supply base structural complexity on financial performance:
Roles of visible and not-so-visible characteristics. Journal of Operations Management, 53:23–44.

Luo, L., Tang, Q., and Lan, Y.-C. (2013). Comparison of propensity for carbon disclosure between
developing and developed countries: A resource constraint perspective. Accounting Research
Journal, 26(1):6–34.

Mann, H. B. and Whitney, D. R. (1947). On a test of whether one of two random variables is

stochastically larger than the other. The Annals of Mathematical Statistics, pages 50–60.

Marshall, D., McCarthy, L., McGrath, P., and Harrigan, F. (2016). What’s your strategy for supply

chain disclosure? MIT Sloan Management Review, 57(2):37–45.

110

McKight, P. E. and Najab, J. (2010). Kruskal-wallis test. The Corsini Encyclopedia of Psychology,

pages 1–1.

McPherson, M., Smith-Lovin, L., and Cook, J. M. (2001). Birds of a feather: Homophily in social

networks. Annual Review of Sociology, 27(1):415–444.

Melnyk, S. A., Schoenherr, T., Speier-Pero, C., Peters, C., Chang, J. F., and Friday, D. (2022). New
challenges in supply chain management: Cybersecurity across the supply chain. International
Journal of Production Research, 60(1):162–183.

Milanesi, M., Runfola, A., and Guercini, S. (2020). Pharmaceutical industry riding the wave of
sustainability: Review and opportunities for future research. Journal of Cleaner Production,
261:121204.

Mitchell, R. K., Agle, B. R., and Wood, D. J. (1997). Toward a theory of stakeholder identification
and salience: Defining the principle of who and what really counts. Academy of Management
Review, 22(4):853–886.

Mollenkopf, D. A., Peinkofer, S. T., and Chu, Y. (2022). Supply chain transparency: Consumer

reactions to incongruent signals. Journal of Operations Management, 68(4):306–327.

Montabon, F., Sroufe, R., and Narasimhan, R. (2007). An examination of corporate reporting,
environmental management practices and firm performance. Journal of Operations Management,
25(5):998–1014.

Moran, P. (2005). Structural vs. relational embeddedness: Social capital and managerial perfor-

mance. Strategic Management Journal, 26(12):1129–1151.

Morosini, P., Shane, S., and Singh, H. (1998). National cultural distance and cross-border acquisi-

tion performance. Journal of International Business Studies, 29(1):137–158.

Nahapiet, J. and Ghoshal, S. (1998). Social capital, intellectual capital, and the organizational

advantage author. Academy of Management Review, 23(2):242–266.

Oracle Corporation (2023). Supply chain transparency defined: Why it matters and its benefits.

Accessed: 2024-05-01.

Pagell, M. and Wu, Z. (2009). Building a more complete theory of sustainable supply chain
management using case studies of 10 exemplars. Journal of Supply Chain Management, 45(2):37–
56.

Pallotti, F. and Lomi, A. (2011). Network influence and organizational performance: The effects
of tie strength and structural equivalence. European Management Journal, 29(5):389–403.

Parsa, S., Roper, I., Muller-Camen, M., and Szigetvari, E. (2018). Have labour practices and

111

human rights disclosures enhanced corporate accountability? The case of the GRI framework.
Accounting Forum, 42(1):47–64.

Patten, D. M. (2002). The relation between environmental performance and environmental disclo-

sure: A research note. Accounting, Organizations and Society, 27(8):763–773.

Pekovic, S., Grolleau, G., and Mzoughi, N. (2018). Environmental investments: Too much of a

good thing? International Journal of Production Economics, 197:297–302.

Phelps, C. C. (2010). A longitudinal study of the influence of alliance network structure and
composition on firm exploratory innovation. Academy of Management Journal, 53(4):890–913.

Podolny, J. M. and Baron, J. N. (1997). Resources and relationships: Social networks and mobility

in the workplace. American Sociological Review, pages 673–693.

Potter, A. and Wilhelm, M. (2020). Exploring supplier–supplier innovations within the Toyota
supply network: A supply network perspective. Journal of Operations Management, 66(7-
8):797–819.

Pucheta-Martínez, M. C. and Gallego-Álvarez, I. (2020). Corporate environmental disclosure
practices in different national contexts: The influence of cultural dimensions. Organization &
Environment, 33(4):597–623.

Ravindran, K., Susarla, A., Mani, D., and Gurbaxani, V. (2015). Social capital and contract
duration in buyer-supplier networks for information technology outsourcing. Information Systems
Research, 26(2):379–397.

Reverte, C. (2009). Determinants of corporate social responsibility disclosure ratings by Spanish

listed firms. Journal of Business Ethics, 88:351–366.

Rezaee, Z. and Tuo, L. (2017). Voluntary disclosure of non-financial information and its association

with sustainability performance. Advances in Accounting, 39:47–59.

Rowley, T., Behrens, D., and Krackhardt, D. (2000). Redundant governance structures: An analysis
of structural and relational embeddedness in the steel and semiconductor industries. Strategic
Management Journal, 21(3):369–386.

Scholtens, B. and Dam, L. (2007). Cultural values and international differences in business ethics.

Journal of Business Ethics, 75:273–284.

Seo, H. (2021). Peer effects in corporate disclosure decisions. Journal of Accounting and Eco-

nomics, 71(1):101364.

Sharma, A., Pathak, S., Borah, S. B., and Adhikary, A. (2020). Is it too complex? the curious case
of supply network complexity and focal firm innovation. Journal of Operations Management,

112

66(7-8):839–865.

Sodhi, M. S. and Tang, C. S. (2018). Corporate social sustainability in supply chains: A thematic

analysis of the literature. International Journal of Production Research, 56(1-2):882–901.

Sodhi, M. S. and Tang, C. S. (2019). Research opportunities in supply chain transparency. Pro-

duction and Operations Management, 28(12):2946–2959.

Song, S., Dong, Y., Kull, T., Carter, C., and Xu, K. (2023). Supply chain leakage of greenhouse gas
emissions and supplier innovation. Production and Operations Management, 32(3):882–903.

Song, S., Lian, J., Skowronski, K., and Yan, T. (2024). Customer base environmental disclosure
and supplier greenhouse gas emissions: A signaling theory perspective. Journal of Operations
Management, 70(3):355–380.

Sousa, R. and Voss, C. A. (2008). Contingency research in operations management practices.

Journal of Operations Management, 26(6):697–713.

Squire, B., Cousins, P. D., and Brown, S. (2009). Cooperation and knowledge transfer within
buyer–supplier relationships: The moderating properties of trust, relationship duration and
supplier performance. British Journal of Management, 20(4):461–477.

Taghizadeh, E., Venkatachalam, S., and Chinnam, R. B. (2021).

on effective resilience assessment of supply networks.
Economics, 241:108254.

Impact of deep-tier visibility
International Journal of Production

Tate, W. L., Dooley, K. J., and Ellram, L. M. (2011). Transaction cost and institutional drivers of

supplier adoption of environmental practices. Journal of Business Logistics, 32(1):6–16.

Tate, W. L., Ellram, L. M., and Bals, L. (2022). Handbook of Theories for Purchasing, Supply

Chain and Management Research. Edward Elgar Publishing.

Tate, W. L., Ellram, L. M., and Gölgeci, I. (2013). Diffusion of environmental business practices:

A network approach. Journal of Purchasing and Supply Management, 19(4):264–275.

Tate, W. L., Ellram, L. M., and Kirchoff, J. F. (2010). Corporate social responsibility reports: A
thematic analysis related to supply chain management. Journal of Supply Chain Management,
46(1):19–44.

Tsai, W. and Ghoshal, S. (1998). Social capital and value creation: The role of intrafirm networks.

Academy of Management Journal, 41(4):464–476.

Uzzi, B. (1996). The sources and consequences of embeddedness for the economic performance

of organizations: The network effect. American Sociological Review, pages 674–698.

113

Uzzi, B. and Lancaster, R. (2003). Relational embeddedness and learning: The case of bank loan

managers and their clients. Management Science, 49(4):383–399.

Van de Ven, A. H. (1976). On the nature, formation, and maintenance of relations among organi-

zations. Academy of Management Review, 1(4):24–36.

Van der Laan Smith, J., Adhikari, A., and Tondkar, R. H. (2005). Exploring differences in social
disclosures internationally: A stakeholder perspective. Journal of Accounting and Public Policy,
24(2):123–151.

Velte, P. (2024). Sustainable board governance and environmental performance: European evi-

dence. Business Strategy and the Environment, 33(4):3397–3421.

Villena, V. H. (2019). The missing link? The strategic role of procurement in building sustainable

supply networks. Production and Operations Management, 28(5):1149–1172.

Villena, V. H. and Craighead, C. W. (2017). On the same page? How asymmetric buyer–supplier
relationships affect opportunism and performance. Production and Operations Management,
26(3):491–508.

Villena, V. H. and Dhanorkar, S. (2020). How institutional pressures and managerial incentives elicit
carbon transparency in global supply chains. Journal of Operations Management, 66(6):697–
734.

Villena, V. H. and Gioia, D. A. (2018). On the riskiness of lower-tier suppliers: Managing

sustainability in supply networks. Journal of Operations Management, 64:65–87.

Villena, V. H., Revilla, E., and Choi, T. Y. (2011). The dark side of buyer–supplier relationships:

A social capital perspective. Journal of Operations Management, 29(6):561–576.

Villena, V. H., Wilhelm, M., and Xiao, C.-Y. (2021). Untangling drivers for supplier environmental
and social responsibility: An investigation in philips lighting’s chinese supply chain. Journal of
Operations Management, 67(4):476–510.

Wang, L., Li, M., Wang, W., Gong, Y., and Xiong, Y. (2023). Green innovation output in the supply
chain network with environmental information disclosure: An empirical analysis of chinese
listed firms. International Journal of Production Economics, 256:108745.

Wang, Y. and Bellemare, M. F. (2019). Lagged variables as instruments. Working paper.

Watts, D. J. and Strogatz, S. H. (1998). Collective dynamics of ‘small-world’ networks. Nature,

393(6684):440–442.

Wilhelm, M., Blome, C., Wieck, E., and Xiao, C. Y. (2016).

Implementing sustainability in
multi-tier supply chains: Strategies and contingencies in managing sub-suppliers. International

114

Journal of Production Economics, 182:196–212.

Wilhelm, M. and Villena, V. H. (2021). Cascading sustainability in multi-tier supply chains: When
do chinese suppliers adopt sustainable procurement? Production and Operations Management,
30(11):4198–4218.

Wooldridge, J. (2007). What’s new in econometrics? lecture 6: Control functions and related
methods. NBER.https://users.nber.org/ confer/2007/si2007/WNE/WNEnotes.pdf, 6:420–445.

Xiao, B., Huang, M., and Barnes, A. J. (2015). Network closure among sellers and buyers in social

commerce community. Electronic Commerce Research and Applications, 14(6):641–653.

Yu, E. P.-Y., Guo, C. Q., and Luu, B. V. (2018). Environmental, social and governance transparency

and firm value. Business Strategy and the Environment, 27(7):987–1004.

Zhao, T. and Wang, H. (2024). The industry peer effect of enterprise ESG performance: The
moderating effect of customer concentration. International Review of Economics & Finance,
92:1499–1525.

Zhu, Q. and Sarkis, J. (2004). Relationships between operational practices and performance among
early adopters of green supply chain management practices in chinese manufacturing enterprises.
Journal of Operations Management, 22(3):265–289.

115

CHAPTER 4

EFFECTIVENESS OF ALTERNATIVE NETWORK
CONFIGURATIONS IN THE AIRLINE INDUSTRY

4.1

Introduction

Operational disruptions, including flight delays, cancellations, and mishandled baggage inci-

dents, can impose substantial costs on airlines and undermine customer satisfaction, ultimately

leading to revenue loss. According to Airlines for America1, delays also cost U.S. air travelers

billions of dollars each year. The Federal Aviation Administration (FAA) estimated that the total

annual cost of delays in the United States reached approximately $33 billion in 2019. This includes

direct costs to airlines, such as increased operational expenses, compensation and assistance to

passengers, as well as indirect costs stemming from lost demand, reputational damage, and broader

economic impacts2. As airlines aim to balance profitability with service quality over the long term,

they are increasingly motivated to improve flight punctuality and mitigate the effects of operational

disruptions.

In today’s fast-paced aviation industry, airlines aim to align with external conditions by strate-

gically reallocating resources to optimize both revenue and operational efficiency (Girod and

Whittington, 2017; Kohl et al., 2007; Miller, 1992). A variety of factors influence on-time per-

formance (OTP) and overall service quality, many of which are tied to operational disruptions.

Research highlights route characteristics and market competition as key drivers of delays and can-

cellations (Prince and Simon, 2015; Rupp and Holmes, 2006; Mazzeo, 2003), alongside operational

and environmental factors such as weather, airport congestion, scheduled block time, and aircraft

size (Deshpande and Arıkan, 2012). Additionally, studies emphasize the interconnected nature of

operational failures, where initial disruptions can cascade and trigger further breakdowns across

airline systems (Parast and Golmohammadi, 2020; Ramdas et al., 2013).

Airlines often improve OTP through various operational strategies, including adding time

buffers to block times, adjusting network structures by adding or removing routes, and implementing

1https://www.airlines.org/dataset/u-s-passenger-carrier-delay-costs/
2https://travelradar.aero/a-data-driven-analysis-of-the-flight-delays-on-airline-profitability/

116

procedural enhancements such as streamlining boarding and reducing gate delays (Manchiraju

et al., 2023). Airline-specific policies, such as baggage fees and boarding procedures, also play

a significant role in shaping operational performance (Nicolae et al., 2017). Ultimately, effective

operational strategies not only enhance efficiency but also support stronger financial performance

(Alan and Lapré, 2018; Ramdas et al., 2013).

Despite the critical importance of managing operational disruptions and maintaining consistent

service quality in the airline industry, prior research has not sufficiently addressed these challenges

from a network perspective. Although preventive strategies to mitigate disruptions continue to

evolve often shaped by the route structures airlines adopt, the impact of different network con-

figurations on airlines’ responses to disruptions remains underexplored. While prior studies have

examined various operational performance metrics in aviation (Manchiraju et al., 2023; Alan and

Lapré, 2018; Phillips and Sertsios, 2013; Tsikriktsis, 2007), comprehensive research linking mul-

tidimensional service quality outcomes to network strategies is still lacking.

The airline network is generally understood to comprise two fundamental components: airports

and the routes that connect them (Ciliberto et al., 2019). Airlines route their flights between

origin and destination airports and form their network structure to optimize their operations. The

two major route systems are point-to-point and hub-and-spoke. In a point-to-point network (i.e.,

decentralized network), the passengers fly directly between airports, whereas, in a hub-and-spoke

network (i.e., centralized network), passengers must make an additional stop through a hub when

neither the origin nor destination airport is a hub (Brueckner, 2004). Figure 4.1 shows two simplified

configurations of widely acknowledged airline networks. Previous studies have compared the two

route systems based on various attributes, including connectivity, the scope of connecting services,

flight frequency, and asset utilization (Zgodavová et al., 2018; Cook and Goodwin, 2008; Oum

et al., 1995).

In response to intensified competition following the introduction of the Airline Deregulation

Act in the U.S. in 1978, the airline industry underwent a major restructuring of carrier networks,

shifting from predominantly linear configurations to the hub-and-spoke model (Mazzeo, 2003; Barla

117

Figure 4.1 Simplified Point-to-Point and Hub-and-Spoke Architectures

and Constantatos, 2000). This structure enables airlines to reallocate capacity more efficiently

across markets by pooling passengers from multiple origins onto a single aircraft (Barla and

Constantatos, 2000). In particular, the hub-and-spoke system offers a distinct operational advantage

through its ability to reallocate capacity ex post under conditions of demand uncertainty (Barla and

Constantatos, 2005, 2000).

While some airlines continue to adopt and invest in point-to-point networks, this structure is

inherently more linear, offering fewer opportunities to consolidate passenger traffic and limiting

the efficiency gains typically associated with hub-based operations. Nevertheless, a point-to-point

network facilitates faster implementation of individual route and origin–destination options, and

enhances organizational adaptability with fewer disruptions by supporting a broader and more

flexible search strategy (Gualini et al., 2023).

U.S. airlines are commonly classified into four categories based on their operational and network

characteristics: Full-Service Carriers (FSCs), Low-Cost Carriers (LCCs), Commuter/Subsidiary

Carriers (CCs), and Leisure Carriers (LCs) (Deshpande and Arıkan, 2012). Among them, American

Airlines, Delta Air Lines, and United Airlines, the three largest FSCs, operate primarily under hub-

and-spoke networks centered around major hubs (Ciliberto et al., 2019; Deshpande and Arıkan,

2012; Tsikriktsis, 2007). In contrast, major LCCs such as Southwest Airlines, JetBlue Airways,

Frontier Airlines, and Spirit Airlines typically adopt point-to-point networks, providing direct

service between distinct city pairs (Zou and Yu, 2020; Ciliberto et al., 2019; Mellat-Parast et al.,

118

2015; Tsikriktsis, 2007). Building on this classification, our study investigates how differences

in network structures across these airline types shape operational performance and service quality

management.

The hub-and-spoke structure commonly employed by FSCs is particularly effective in markets

that demand high flight frequency, as it supports extensive service coverage while maintaining ac-

ceptable total travel times for connecting passengers (Brueckner, 2004; Mayer and Sinai, 2003). The

superior rerouting capabilities of FSCs also enable them to mitigate the impact of delays by provid-

ing alternative flight paths, thereby illustrating how network design can improve key performance

metrics through the strategic deployment of resources (Mellat-Parast et al., 2015). However, this

model also entails high operational demands and increased vulnerability to congestion, particularly

at major hub airports, rendering FSCs more susceptible to disruptions such as delays in the event

of hub failures (Ding et al., 2023; Bauranov et al., 2021; Miller, 1992). Nonetheless, direct flights

serving major airports tend to recover more swiftly from operational disruptions, owing to higher

passenger demand, enhanced network connectivity, and greater availability of operational resources

(Sugishita et al., 2024). In contrast, LCCs benefit from lower operating costs by avoiding complex

connection schedules, utilizing less congested secondary airports with reduced fees, and achieving

quicker aircraft turnaround times (Bitzan and Peoples, 2016). However, the point-to-point network

structure typically employed by LCCs is more linear, offering limited opportunities for passenger

pooling and thereby constraining the scale economies associated with hub-based operations. While

LCCs are generally less susceptible to delays associated with hub airport disruptions and benefit

from greater operational flexibility due to their standardized aircraft fleets, their low-cost model

limits the availability of redundant resources, making recovery from disruptions more challenging.

Figures 4.2 and 4.3 reveal clear distinctions between FSCs and LCCs in 2019 across two key

operational metrics: OTP and the average number of mishandled bags per 1,000 enplaned bags.

Specifically, Figure 4.2 shows that both carrier types follow similar seasonal trends in OTP, though

FSCs consistently exhibit a slight performance advantage.

In contrast, Figure 4.3 reveals that

FSCs consistently underperform LCCs in minimizing mishandled baggage incidents. While these

119

Figure 4.2 On-time Performance in 2019

Figure 4.3 Average Reports per 1,000 Bags Enplaned in 2019

120

patterns suggest performance differences between network types, it remains unclear whether these

differences are statistically significant or systematic across various metrics. Moreover, it is not

immediately evident which carrier type performs better on which specific dimension of service

quality. This ambiguity underscores the need for a systematic empirical investigation and forms

the basis for the research questions addressed in this study.

4.1.1 Research Questions

The primary objective of this study is to investigate how airlines employing different network

structures manage service quality, as evidenced by standard operational performance metrics widely

used in the industry. Despite the importance of this topic, prior research has not adequately examined

how service quality management practices vary across carriers with distinct network strategies.

To address this gap, we conduct a comprehensive analysis and comparison of key performance

indicators (KPIs) that quantify operational performance under disruptions in the airline industry.

Specifically, we examine five metrics: (1) departure delays, (2) arrival delays, (3) OTP, (4) flight

cancellations, and (5) baggage handling incidents, comparing outcomes between FSCs and LCCs.

We assess whether operational and service quality outcomes systematically differ based on the

network configuration adopted by an airline.

Moreover, we investigate whether there are significant differences in operational performance

and service quality management among the group of LCCs. This question is particularly relevant

in the context of Southwest Airlines, one of the largest U.S. carriers, which also operates a point-to-

point network, albeit with a stronger presence and greater operational influence at specific airports

within its network. Given both the operational scale and unique characteristics of Southwest, we

seek to examine the extent to which variations in operations and service quality management exist

within the broader LCC category.

4.1.2 Our Results and Contributions

Our empirical analysis draws on historical operational and service performance data from seven

major U.S. domestic carriers, revealing systematic differences in performance across airline types

based on five key metrics: departure and arrival delays, OTP, flight cancellations, and mishandled

121

baggage incidents. First, we find that FSCs consistently outperform LCCs in minimizing both

departure and arrival delays. Notably, within the LCC group, Southwest Airlines achieves lower

arrival delays than its peers. Second, our findings indicate that FSCs are significantly more effective

in maintaining high OTP and reducing cancellations. Specifically, FSCs have 53.9% higher odds

of achieving on-time arrivals and 37% lower odds of cancellations compared to LCCs. While

Southwest’s OTP is statistically comparable to that of other LCCs, it performs worse than both

FSCs and its LCC counterparts in managing cancellations. Finally, LCCs show better performance

in baggage handling compared to FSCs, with Southwest outperforming all carriers in minimizing

the number of mishandled bags per 1,000 enplaned bags.

Unlike prior studies that focus on a limited set of performance indicators, our study contributes

to the literature by systematically comparing multiple dimensions of operational performance across

different types of carriers. Using historical flight-level data from several U.S. airlines, we develop

a comprehensive framework to assess how differences in network design influence service quality

and operational reliability. Our findings suggest that network configuration plays a critical role in

shaping performance outcomes, highlighting the strategic importance of network design decisions

for airline operations. These insights underscore the need for airline managers to carefully con-

sider their network structure when evaluating and benchmarking operational performance against

competitors.

4.1.3 Organization

Our paper is organized as follows. Section 2 reviews the relevant literature on airline net-

work structures and operational strategies, with particular attention to risk mitigation approaches

discussed in both the aviation and supply chain management literatures. Section 3 presents our

hypotheses, developed through a detailed examination of commonly used operational performance

metrics, and introduces a set of competing hypotheses grounded in prior research. Section 4 outlines

the research design, including data sources, summary statistics, and variable definitions. Section 5

describes the sample selection process, the matching procedure used to ensure comparable units of

analysis, and the estimation models employed, followed by a discussion of the empirical results. In

122

Section 6, we conduct a separate analysis of Southwest Airlines, distinguishing it from other LCCs

to explore within-group differences based on our network-based classification. Finally, Section 7

concludes by discussing the implications of our findings and offering directions for future research.

4.2 Literature Review

We position our study within the broader academic context by reviewing the existing literature

related to airline operational performance and network configurations.

4.2.1 Managing Operational Disruptions and Performance in the Airline Industry

Understanding the implications of airline network structures is important when analyzing oper-

ational disruptions in the airline industry such as flight delays, flight cancellations, and mishandled

baggage incidents. Among these, flight delays can be broadly categorized into two components:

intrinsic delays, which arise from factors specific to the flight itself, and propagated delays, which

result from spillover effects across an aircraft’s rotation cycle (Nicolae et al., 2017; Arıkan et al.,

2013). Intrinsic delay is typically measured by comparing the intrinsic duration of a flight (i.e.,

actual flight duration excluding spillover delays from previous flights) to its scheduled duration,

calculated as the difference between scheduled departure and arrival times. The two critical dimen-

sions of delays, intrinsic and propagated, play a central role in shaping overall service performance.

Notably, flight arrival delays, a primary contributor to system-wide disruptions, are often influenced

by a combination of intrinsic and propagated delays (Malladi and Sohoni, 2022). Accordingly, op-

erational failures in the airline industry are frequently assessed using performance indicators such

as flight delays, OTP, flight cancellations, and mishandled baggage rates, each of which serves as

a proxy for service quality and reliability (Alan and Lapré, 2018; Deshpande and Arıkan, 2012;

Tsikriktsis, 2007).

Prior studies have identified a range of factors contributing to operational disruptions, including

adverse weather, airport congestion, aircraft utilization, market competition, and route character-

istics (Deshpande and Arıkan, 2012; Rupp and Holmes, 2006; Mazzeo, 2003). These disruptions

impose significant costs, such as crew overtime, increased fuel consumption, and passenger re-

accommodation expenses, prompting airlines to continually refine their operational strategies to

123

mitigate financial and reputational risks (Hassan et al., 2021). To improve OTP and reduce flight

cancellations, airlines adopt strategic measures such as optimizing flight schedules, improving gate

management, and adjusting network configurations (Prince and Simon, 2015, 2009). In addition,

streamlined operations and efficient fleet utilization shaped by factors like flight demand, aircraft

age, and maintenance capacity are closely linked to enhanced profitability and service quality

(Tsikriktsis, 2007).

Despite advancements in the literature, several important limitations remain. Most existing

studies focus on a single outcome variable, such as flight delays or cancellations, while rarely

considering multiple dimensions of service quality in tandem. Furthermore, the influence of

airline network structure, particularly the distinction between hub-and-spoke and point-to-point

configurations, remains underexamined in relation to operational performance. Although prior

research suggests that point-to-point networks reduce delay propagation due to route independence

(Cook and Goodwin, 2008), there is limited empirical work that systematically compares how

different network structures shape service outcomes. Furthermore, the choice and interpretation

of operational metrics, such as measuring mishandled baggage per passenger versus per enplaned

bag, can significantly influence assessments of airline performance. To address these gaps, this

study jointly examines five key operational performance metrics: departure delays, arrival delays,

OTP, flight cancellations, and mishandled baggage incidents. By adopting a multidimensional

approach, this study provides a more comprehensive understanding of how network configurations

and operational strategies affect airline service quality and reliability.

4.2.2 Airline Operations and Network Design

Most FSCs operate under a hub-and-spoke network model, channeling a substantial portion of

passenger traffic through centralized hub airports. This configuration streamlines operations by

enabling airlines to serve numerous destinations with fewer direct routes, while also enhancing

resource efficiency through optimized aircraft deployment, crew scheduling, and airport utilization

(Cook and Goodwin, 2008). The model improves connectivity across the network and allows

airlines to adjust capacity more effectively, for instance, by deploying larger aircraft on high-

124

Table 4.1 Comparing Hub-and-Spoke and Point-to-Point Networks

Operational Metrics

Hub-and-Spoke

Point-to-Point

Number of routes needed to
connect all nodes in a network

Lower number of routes

Higher number of routes

Market Size

Efficiently serves cities of
widely varying sizes

Depends on high-density markets
with one high-demand endpoint

Frequency of flights

Higher frequency

Lower frequency

Connectivity

Hub connections
for continuing flights

No connecting flights

Size of aircraft fleet

Small aircraft fleet

Large aircraft fleet

Fleet type

Wide variation
in seating capacity

Single fleet type

Total travel time

Longer travel time

Shorter travel time

demand routes between major hubs or international destinations, thereby minimizing the number

of flights and total seat-kilometers offered (Zgodavová et al., 2018).

In contrast, LCCs favor point-to-point networks that offer direct service between origin and

destination airports (Parast and Golmohammadi, 2020; Zou and Yu, 2020; Nicolae et al., 2017;

Mellat-Parast et al., 2015; Lordan, 2014; Oum et al., 1995). Unlike the hub-and-spoke model, this

approach requires a greater number of routes to expand network coverage. For instance, an airline

operating from five airports would need 10 direct routes; adding a sixth destination would require

five more routes to maintain full connectivity. While this model enhances flexibility and reduces

reliance on centralized hubs, it often results in lower passenger load factors and demands significant

investment in fleet size and personnel. Overall, each air transport network (ATN) exhibits distinct

operational characteristics, as summarized in Table 4.1.

FSCs generally support longer aircraft turnaround times to coordinate feeder and trunk routes,

thereby enhancing load factors and benefiting from economies of traffic density. This structure is

particularly advantageous in markets requiring high flight frequency, as it maintains broad service

125

coverage without significantly compromising total travel time for connecting passengers (Brueckner,

2004; Mayer and Sinai, 2003). Conversely, LCCs are designed for faster aircraft turnaround at the

gate (Arıkan et al., 2013), allowing efficient operations even with smaller passenger volumes

(Gualini et al., 2023). Thus, while hub-and-spoke networks optimize capacity through strategic

hub placement, albeit with longer turnaround times, point-to-point networks are favored for their

operational efficiency under tightly coupled schedules requiring minimal buffer time.

Although understanding the impact of the network structure of airlines on their operational per-

formance is important, empirical evidence to understand network centrality and airline operational

performance remains limited. This study, therefore, aims to understand the impact of different

network structures on key operational performance metrics in the U.S. airline industry.

4.2.3 Flexibility vs. Redundancy

Our study also relates to the concept of resilience, which refers to the ability to recover from

disruptions and is commonly examined within the risk management literature. In the context of

supply chain risk management, potential risks include, but are not limited to, delays, disruptions,

forecast inaccuracies, system breakdowns, and capacity issues (Sodhi and Chopra, 2004). When

disruptions happen, firms implement diverse risk mitigation strategies to recover and restore oper-

ations. Previous literature suggests that firms can achieve resilience by either increasing flexibility

or creating redundancy, leading to the identification of these two primary risk mitigation strategies,

redundancy and flexibility, as key methods for strengthening organizational resilience (Talluri et al.,

2013; Sheffi and Rice Jr, 2005).

Reducing vulnerability is basically a corporate strategic initiative aimed at increasing resilience,

which can be achieved by either creating redundancy or enhancing flexibility (Sheffi and Rice Jr,

2005). While redundancy is a component of every resilience strategy, it often incurs substantial

costs with limited benefits. In contrast, flexibility can provide a competitive advantage in daily

operations (Sheffi and Rice Jr, 2005). Focusing on the manufacturing sector, Talluri et al. (2013)

evaluate seven risk mitigation strategies across nine potential failure types and suggest that the

most effective approaches do not necessarily involve shielding firms from disruptions (e.g., through

126

redundancy), but rather emphasize enhancing recovery capabilities (e.g., through flexibility).

In the commercial aviation industry, a variety of factors shape airlines’ ability to respond to

disruptive events, as carriers continuously strive to improve recovery from operational disruptions

(Park et al., 2018; Kohl et al., 2007). To maintain service quality amid uncertainty, airlines

often reallocate internal resources to manage rising operational costs and volatility caused by such

disruptions (Hassan et al., 2021). An airline’s capacity to substitute aircraft and crew as needed

plays a critical role in its ability to respond effectively, and this is largely determined by flight

scheduling practices, aircraft assignments, and crew availability (Ramdas and Williams, 2006).

Securing sufficient capacity enables more efficient resource reallocation, ultimately enhancing

resilience in the face of service failures.

As shown in Table 4.1, the hub-and-spoke configuration, commonly used by FSCs, relies on

centralized hubs and connecting flights, which complicates operational adjustments during disrup-

tions and reduces flexibility. To support both international and domestic services, FSCs also operate

diverse aircraft fleets, further increasing scheduling and maintenance complexity. However, this

network structure enables more efficient resource allocation and operational efficiency, supported

by economies of scale and greater resource availability. FSCs’ superior rerouting capabilities allow

them to minimize the impact of delays by offering alternative flight paths, highlighting how network

design can enhance key performance metrics through effective resource deployment (Mellat-Parast

et al., 2015).

In contrast to hub-and-spoke systems, point-to-point networks, commonly used by LCCs,

offer greater operational flexibility, despite limited redundancy. Operating on tight schedules

with minimal spare capacity, LCCs are less able to absorb disruptions. However, their network

structure avoids intermediate stops and relies on more decentralized networks. Their point-to-point

flights reduce total travel time while maintaining route independence (Cook and Goodwin, 2008).

Further, as shown in Table 4.1, LCCs typically use homogeneous fleets, enabling more efficient

crew assignments, streamlined maintenance, and enhanced flexibility.

In summary, FSCs excel in redundancy, offering multiple backup systems and service layers

127

that help mitigate the impact of operational disruptions. In contrast, LCCs demonstrate superior

flexibility, particularly in terms of operational and network responsiveness, enabling them to

adapt more rapidly to disruptions or changing market and environmental conditions. Our study

contributes to the risk management literature by evaluating the relative effectiveness of redundancy

and flexibility in managing disruptions within service networks, with a particular emphasis on air

transportation systems.

4.2.4 Organizational Structure

It has been acknowledged that the design and structure of an organization have a significant

impact on the performance and effectiveness of the organization (Burton and Obel, 2018; Dalton

et al., 1980; Van de Ven, 1976; Campbell et al., 1974). Organizational design prescribes the

optimal structure for an organization to function effectively and efficiently (Burton and Obel, 2018;

Dalton et al., 1980). Organizational structure is often defined as “the enduring characteristics of an

organization, reflected in the arrangement of units and positions and their systematic relationships

to one another” (James and Jones, 1976). It addresses various organizational attributes including

specialization, centralization, standardization, and formalization (Campbell et al., 1974). The

organizational structure serves two primary functions: 1) minimizing or regulating the impact of

individual variations on the organization, and 2) exerting power to enable decision-making and the

execution of organizational activities (Dalton et al., 1980).

The extensive academic literature explores the impact of organizational structure on organiza-

tional performance (Dalton et al., 1980; Van de Ven, 1976). Specifically, the extent of centralization,

which reflects the concentration of decision-making authority among a few entities within an orga-

nization, is widely recognized as a critical factor influencing organizational performance. Previous

literature suggests that learning spillover effects can be amplified through centralization by pro-

moting the adoption of common practices (Chang and Harrington, 2000). On the other hand,

it is frequently suggested that uncertainty, which affects decision-making processes, calls for the

delegation of control through decentralization to ease the managerial workload (Miller, 1992).

Nonetheless, empirical evidence on the relationship between centralization and organizational

128

performance remains limited, particularly within the service sector.

4.3 Hypotheses Development

In this section, we develop hypotheses to assess the comparative performance of point-to-point

and hub-and-spoke airline network structures across five key operational metrics that quantify

performance under disruptions in the airline industry: departure delays, arrival delays, OTP, flight

cancellations, and mishandled baggage incidents.

4.3.1 Departure and Arrival Delays

In service operations, departure and arrival delays serve as key indicators of service quality

and operational consistency (Mellat-Parast et al., 2015; Ramdas et al., 2013; Tsikriktsis, 2007).

Airlines can reduce delays by strategically balancing capacity and inventory, and by maintaining

flexible excess capacity (Sodhi and Chopra, 2004). ATNs critically influence both service quality

and operational profitability (Oum et al., 1995). Point-to-point networks, commonly used by LCCs,

reduce reliance on hub airports by operating independent direct flights. This structure minimizes

delay propagation, insulates the broader schedule from localized disruptions, and reduces total

travel time by eliminating intermediate stops (Cook and Goodwin, 2008). Combined with fuel-

efficient aircraft and operational simplicity, point-to-point systems are increasingly favored for

their resilience and flexibility in handling disruptions (Kohl et al., 2007). Moreover, LCCs often

operate homogeneous fleets, allowing for streamlined operations in which crew and aircraft can be

easily reassigned, thereby enhancing scheduling efficiency, reducing maintenance complexity, and

increasing overall flexibility.

In contrast, the hub-and-spoke model, typically adopted by FSCs, offers advantages in connec-

tivity, market coverage, and capacity optimization. Despite greater vulnerability to systemic delays

due to its centralized structure, this model supports higher flight frequencies and enables strategic

capacity allocation through passenger pooling (Zgodavová et al., 2018; Cook and Goodwin, 2008).

FSCs also maintain ample backup and slack resources to ensure smooth operations, operate diverse

fleets, provide differentiated services, and achieve economies of scale by concentrating traffic on

fewer, high-frequency routes (Parast and Golmohammadi, 2020; Tsikriktsis, 2007). While less

129

robust to cascading disruptions, the model enhances operational efficiency on a larger scale.

As defined by the U.S. Bureau of Transportation Statistics (BTS), a departure delay refers to the

number of minutes between a flight’s scheduled and actual departure times, while an arrival delay

captures the gap between scheduled and actual arrival times. These delays can propagate across

an aircraft’s rotation cycle, leading to spillover effects that disrupt subsequent flights (Nicolae

et al., 2017; Arıkan et al., 2013). Although arrival delays frequently lead to departure delays,

prior research highlights a gap in how airlines manage the two, owing to various operational and

contingency factors such as the national aviation system, hub connectivity, weather conditions, and

crew availability (Parast and Golmohammadi, 2020; Ramdas and Williams, 2006; Mayer and Sinai,

2003). While FSCs are often viewed as more capable of managing disruptions due to their ample

and redundant resource base, LCCs may naturally reduce delay risks through simplified operations.

These contrasting strengths and limitations in managing different types of delays motivate the

following set of competing hypotheses for both departure and arrival delays.

Hypothesis 1A (H1A) Full-service carriers utilizing hub-and-spoke networks outperform low-

cost carriers utilizing point-to-point networks in managing departure delays.

Hypothesis 1B (H1B) -competing. Low-cost carriers utilizing point-to-point networks outper-

form full-service carriers utilizing hub-and-spoke networks in managing departure delays.

Hypothesis 1C (H1C) Full-service carriers utilizing hub-and-spoke networks outperform low-

cost carriers utilizing point-to-point networks in managing arrival delays.

Hypothesis 1D (H1D) -competing. Low-cost carriers utilizing point-to-point networks outper-

form full-service carriers utilizing hub-and-spoke networks in managing arrival delays.

4.3.2 On-time Performance

OTP is widely recognized as a key indicator of delivery reliability and overall service quality in

the airline industry (Manchiraju et al., 2023; Alan and Lapré, 2018; Tsikriktsis, 2007). Airlines can

improve OTP by strategically allocating critical resources, such as personnel, aircraft, and flight

schedules, to manage disruptions and maintain punctuality (Manchiraju et al., 2023; Prince and

130

Simon, 2015, 2009). A key determinant of OTP is buffer time, which includes both block-time (the

difference between scheduled and actual flight time) and ground-time buffers. These buffers help

absorb operational variability and prevent delay propagation.

However, airlines face a trade-off: while longer buffers enhance schedule reliability and improve

OTP, they reduce efficiency by lowering aircraft utilization and increasing turnaround times. This

tension is particularly pronounced across different network structures. Point-to-point networks,

which emphasize high utilization and direct services, minimize buffers and are therefore more

susceptible to OTP deterioration under delays. In contrast, hub-and-spoke networks include more

slack, especially around hub connections, offering greater resilience but exposing operations to

congestion and coordination challenges.

Prior research shows that airline network structures significantly influence OTP outcomes by

shaping buffer strategies and exposure to operational disruptions (Arıkan et al., 2013). Flights from

major hubs often experience lower OTP due to congestion, connecting passenger coordination, and

concentrated inbound traffic (Deshpande and Arıkan, 2012; Mayer and Sinai, 2003). However,

other studies suggest that major hubs may offer more reliable connections than smaller airports

due to superior infrastructure and resource availability (PeCoy and Redmond, 2023; Redmond

et al., 2019). Airlines operating on less competitive routes may also have weaker incentives to

maintain punctuality, regardless of network type (Prince and Simon, 2015, 2009; Mazzeo, 2003).

While airlines attribute flight delays to factors beyond their control, research shows that they are

shaped by strategic and operational decisions. Higher OTP among LCCs may reflect their reduced

exposure to hub congestion, whereas FSCs often invest more heavily in mitigating delay-related

costs to maintain schedule reliability. FSCs may therefore be better equipped to manage OTP under

complex conditions, while LCCs benefit from streamlined operations and shorter turnaround times

that naturally support punctuality. Taken together, these observations suggest that both FSCs and

LCCs can achieve strong OTP performance, though through distinct mechanisms: FSCs leverage

operational redundancy, resource efficiency, and built-in schedule buffers; LCCs operate through

simplified, tightly scheduled, decentralized networks that emphasize flexibility. Based on this

131

reasoning, we propose the following set of competing hypotheses:

Hypothesis 2A (H2A) Full-service carriers utilizing hub-and-spoke networks outperform low-

cost carriers utilizing point-to-point networks in maintaining on-time performance.

Hypothesis 2B (H2B) -competing. Low-cost carriers utilizing point-to-point networks outper-

form full-service carriers utilizing hub-and-spoke networks in maintaining on-time performance.

4.3.3 Managing Flight Cancellations

According to the U.S. BTS, the primary causes of flight cancellations include extreme weather,

operational issues (e.g., mechanical problems, crew shortages, late-arriving aircraft, and high traffic

volume), and security concerns such as terminal evacuations or re-boarding after security breaches3.

While cancellations can also result from low passenger demand or economic considerations, as

airlines weigh the costs of operating a flight against canceling it (Alderighi and Gaggero, 2018; Rupp

et al., 2006), they are often used strategically to minimize delays and contain irregular operations.

Consequently, when faced with disruptions such as cancellations, congestion, or delays, airlines

must adapt their operational strategies to maintain service quality and avoid financial strain.

Previous research shows that flight cancellations are shaped by both network structure and

market conditions. Routes with high traffic concentration and operations at congested airports are

more prone to delays and cancellations (Alderighi and Gaggero, 2018; Rupp et al., 2006; Mazzeo,

2003). LCCs, which rely on decentralized point-to-point networks with lower flight frequencies,

generally report lower cancellation rates due to reduced dependency on individual airports and

lower risk of cascading disruptions (Alderighi and Gaggero, 2018). Prior studies also suggest that

routes with lower traffic concentration are less vulnerable to congestion, resulting in fewer delays

and cancellations (Alderighi and Gaggero, 2018; Rupp et al., 2006; Mazzeo, 2003). Differing

from this pattern, FSCs operating hub-and-spoke systems benefit from resource concentration at

hubs, enabling faster recovery and aircraft substitution (Kohl et al., 2007; Rupp et al., 2006). Hub

airlines are also less likely to cancel flights given the risk of disrupting multiple connections (Mayer

and Sinai, 2003). However, alliance-affiliated FSCs often face higher cancellation rates overall,

3https://www.bts.gov/topics/airlines-and-airports/understanding-reporting-causes-flight-delays-and-cancellations

132

largely due to congestion and operational complexity at large hubs (Alderighi and Gaggero, 2018).

Beyond network design, cancellation rates also rise with market concentration, reflecting reduced

service quality under weaker competition (Alderighi and Gaggero, 2018; Cao et al., 2017; Rupp

and Holmes, 2006), while flexible scheduling strategies, such as buffer times and idle capacity,

can mitigate cancellations (Ramdas et al., 2013; Barnhart et al., 2012). These contrasting findings

highlight that both FSCs and LCCs face distinct trade-offs in managing cancellations, motivating

the following set of competing hypotheses.

Hypothesis 3A (H3A) Full-service carriers utilizing hub-and-spoke networks outperform low-

cost carriers utilizing point-to-point networks in minimizing flight cancellations.

Hypothesis 3B (H3B) -competing. Low-cost carriers utilizing point-to-point networks outper-

form full-service carriers utilizing hub-and-spoke networks in minimizing flight cancellations.

4.3.4 Mishandled Baggage Incidents

In recent years, the rapid growth of the air travel industry and rising passenger volumes have led

to an overall increase in the total number of mishandled bags, despite improvements in mishandled

baggage rates. According to Société Internationale de Télécommunications Aéronautiques (SITA),

more than 10 million additional bags were delayed, lost, misdirected, pilfered, or stolen in 2024

compared to previous years4. In the airline industry, average service quality is typically assessed

using indicators such as OTP and mishandled baggage rates; the latter reflects an airlines’ ability to

recover lost items efficiently, which is a key component of service quality given the additional costs

of locating and delivering delayed baggage, and customer dissatisfaction associated with baggage

recovery (Alan and Lapré, 2018; Mellat-Parast et al., 2015; Phillips and Sertsios, 2013; Tsikriktsis,

2007).

Airports designated as hubs or focus cities introduce added operational complexity, largely due

to the need to transfer baggage between connecting flights (Nicolae et al., 2017). This complexity

is further amplified when airlines operate through large primary airports, such as Chicago O’Hare

and Dallas/Fort Worth, compared to secondary airports like Chicago Midway or Dallas Love Field,

4https://www.sita.aero/resources/surveys-reports/sita-baggage-it-insights-2024/

133

which are less congested and typically expose airlines to fewer service failures (Alan and Lapré,

2018). From a network design perspective, these complexities are more pronounced for FSCs,

which face tighter scheduling constraints and manage more diverse fleets than LCCs (Alan and

Lapré, 2018). Moreover, FSCs facilitate a higher volume of connecting flights, increasing customer

contact points and baggage handling requirements, thereby elevating the risk of service disruptions

(Alan and Lapré, 2018). In contrast, the point-to-point network structure commonly adopted by

LCCs reduces the likelihood of baggage mishandling, as passengers typically reclaim their luggage

at each destination. Reflecting these operational differences, Parast and Golmohammadi (2020)

find that FSCs, based on data from 1998 to 2009, report lower overall service quality, as measured

by customer complaints, compared to LCCs.

When service failures occur, airlines must mobilize significant resources such as frontline staff

time and compensation to manage disruptions (Alan and Lapré, 2018). While larger airlines face

greater organizational complexity, they also benefit from more operational slack and resource

availability, enabling more effective recovery (Hassan et al., 2021; Parast and Golmohammadi,

2020). FSCs, supported by greater operational slack and more robust baggage handling systems,

may be better positioned to manage incidents such as mishandled baggage. Based on this reasoning,

we propose the following set of competing hypotheses:

Hypothesis 4A (H4A) Full-service carriers utilizing hub-and-spoke networks outperform low-

cost carriers utilizing point-to-point networks in reducing mishandled baggage incidents.

Hypothesis 4B (H4B) -competing. Low-cost carriers utilizing point-to-point networks out-

perform full-service carriers utilizing hub-and-spoke networks in reducing mishandled baggage

incidents.

4.4 Data

In this section, we discuss the sources of our data sets and variables used for empirical analysis.

An overview of our data sources and variables is provided in Table 4.2.

134

4.4.1 Databases

We obtained the data for our empirical analysis from the official websites of the BTS and the

U.S. Department of Transportation (DOT). Consistent with prior research on operational perfor-

mance in the U.S. airline industry (Manchiraju et al., 2023; Alan and Lapré, 2018; Nicolae et al.,

2017; Deshpande and Arıkan, 2012), we primarily draw on the OTP dataset, which includes de-

tailed flight-level records for all U.S. air carriers accounting for at least one percent of domestic

scheduled passenger revenue. This dataset provides comprehensive operational information for

each commercially scheduled flight operated by these carriers.

For the purpose of hypothesis testing, we focus on OTP data spanning from 2016 to 2019. This

period offers comprehensive coverage of key variables used in our empirical models, including

carrier identifiers, origin and destination airports, departure and arrival performance metrics, as

well as flight cancellations and diversions. These variables form the foundation of both the

dependent and independent constructs examined in this study. As such, the OTP dataset serves as

a central resource for our analysis.

In addition, we collect mishandled baggage data from Air Travel Consumer Reports provided by

U.S. DOT. This report is published monthly by the DOT’s Office of Aviation Consumer Protection

and is intended to provide consumers with information on the quality of services offered by airlines5.

According to the U.S. DOT, baggage statistics are based on data reported by U.S. air carriers that

account for at least one-half of one percent of total domestic scheduled-service passenger revenues,

as determined by DOT’s BTS. According to the monthly Air Travel Consumer Reports, the number

of mishandled bags is defined as “the number of checked bags that are lost, damaged, delayed, or

pilfered, as reported by or on behalf of the passenger.”

Beginning in January 2019, the U.S. DOT revised its reporting criteria for mishandled baggage

in the Air Travel Consumer Reports. Prior to this change, as noted in several previous studies

(Alan and Lapré, 2018; Mellat-Parast et al., 2015; Tsikriktsis, 2007), mishandled baggage was

reported as the number of lost, damaged, delayed, or pilfered bags per 1,000 passengers. However,

5https://www.transportation.gov/individuals/aviation-consumer-protection/air-travel-consumer-reports

135

since the revision, the metric has been updated to reflect the number of mishandled bags per 1,000

enplaned bags, which serves as a key indicator of airline service quality. To more accurately assess

the relative performance of baggage handling across airlines, we adopt this updated metric in our

analysis. Accordingly, we use mishandled baggage data covering the period from January 2019

to February 2020. This timeframe is selected to minimize the influence of external disruptions,

concluding just prior to the World Health Organization’s (WHO) declaration of COVID-19 as a

global pandemic on March 11, 2020.6

4.4.2 Variables

We begin our empirical analysis by defining the dependent and independent variables used to

test our hypotheses.

4.4.2.1 Dependent Variables

To test the first two sets of competing hypotheses, we use both departure and arrival delays

in this study. Departure delays and arrival delays have been used to measure airline operational

performance in previous studies (Nicolae et al., 2017; Deshpande and Arıkan, 2012; Ramdas and

Williams, 2006). As defined by the BTS, a departure delay is defined as the difference in minutes

between a flight’s scheduled and actual departure times, whereas an arrival delay is defined as the

difference in minutes between a flight’s scheduled and actual arrival times. Thus, we define two

dependent variables to test the first two sets of hypotheses: 𝐷𝑒 𝑝_𝐷𝑒𝑙𝑎𝑦𝑖𝑡𝑘 and 𝐴𝑟𝑟_𝐷𝑒𝑙𝑎𝑦𝑖𝑡𝑘 .

Specifically, 𝐷𝑒 𝑝_𝐷𝑒𝑙𝑎𝑦𝑖𝑡𝑘 (resp., 𝐴𝑟𝑟_𝐷𝑒𝑙𝑎𝑦𝑖𝑡𝑘 ) represents the departure (resp., arrival) delay

of a flight operated by a carrier 𝑖, at time 𝑡, on route 𝑘.

Several metrics are utilized in the airline industry to assess operational performance. Foremost

among these are OTP and late arrivals, which are typically regarded as key internal measures

of punctuality and service reliability (Manchiraju et al., 2023; Alan and Lapré, 2018; Tsikriktsis,

2007). According to the U.S. DOT, a flight is classified as delayed if it arrives at the destination

gate 15 minutes or more after the scheduled arrival time (Ramdas et al., 2013; Deshpande and

6https://www.who.int/director-general/speeches/detail/who-director-general-s-opening-remarks-at-the-media-

briefing-on-covid-19—11-march-2020

136

Arıkan, 2012). Similarly, OTP is defined as the proportion of flights of an airline that arrive within

15 minutes of their scheduled arrival time (Alan and Lapré, 2018). Based on this, to test the third

set of competing hypotheses, we construct a dummy variable 𝑂𝑇 𝑃𝑖𝑡𝑘 as follows:

𝑂𝑇 𝑃𝑖𝑡𝑘 =






1,

if the flight operated by airline 𝑖, at time 𝑡, on route 𝑘, arrives within

15 minutes of its scheduled arrival time,

0,

otherwise.

We consider the proportion of flights canceled as the fourth dependent variable. Ramdas et al.

(2013) suggest that cancelled flights often lead to prolonged delays and the added inconvenience of

rebooking for travelers. Consequently, such extreme disruptions caused by cancellations are more

likely to have a lasting negative impact on customers’ perceptions of an airline and influence their

future purchasing decisions (Ramdas et al., 2013). Accordingly, we incorporate flight cancellations

as another indicator of service quality. We operationalize flight cancellations using the variable

𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛𝑖𝑡𝑘 as follows:

𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛𝑖𝑡𝑘 =

1,

0,





if the flight operated by airline 𝑖, at time 𝑡, on route 𝑘, is cancelled,

otherwise.

Finally, we use the number of mishandled bags per 1,000 bags enplaned as the last dependent

variable in this study. Previous studies often rely on the number of mishandled baggage per 1,000

passengers to measure average service quality (Alan and Lapré, 2018; Nicolae et al., 2017; Mellat-

Parast et al., 2015; Phillips and Sertsios, 2013). However, variations in free checked baggage

policies may distort the evaluation of airline service quality. For instance, Southwest Airlines’

“Bags Fly Free” policy is a central component of its marketing strategy and value proposition, but

this policy may result in an increase in the volume of checked baggage (Nicolae et al., 2017). As

a result, using the number of mishandled bags per 1,000 passengers as a service quality metric

could disproportionately penalize Southwest, as the airline processes a higher volume of checked

137

baggage, increasing the likelihood of baggage mishandling incidents. To account for potential

distortions arising from differences in checked baggage policies across carriers, we use the number

of mishandled bags per 1,000 enplaned bags as our dependent variable. Specifically, we define

𝐵𝑎𝑔_𝐼𝑠𝑠𝑢𝑒 𝑗 𝑦𝑚 as the number of bags mishandled per 1000 bags enplaned by carrier 𝑗 in month 𝑚

of year 𝑦.

4.4.2.2 Explanatory Variables

In this study, we use two main independent variables to test our hypotheses. The first independent

variable classifies each carrier as either an FSC or an LCC. First, we categorize American Airlines

(AA), Delta Air Lines (DL), and United Airlines (UA) as FSCs, while JetBlue Airways (B6),

Frontier Airlines (F9), Spirit Airlines (NK), and Southwest Airlines (WN) as LCCs7. We define

the time-invariant dummy variable 𝐹𝑆𝐶 𝑗 to distinguish between FSCs and LCCs as follows:

𝐹𝑆𝐶 𝑗 =

1,

0,





if the airlines is a full-service carrier,

otherwise.

Beyond the binary classification represented by 𝐹𝑆𝐶 𝑗 , we refine the carrier type variable into

a categorical form by distinguishing Southwest Airlines as its own category in this study. Since

Southwest Airlines operates a point-to-point network but with a stronger influence at specific

airports within its network (e.g., DAL, HOU, and MDW airports), its network design allows it to

achieve both flexibility and redundancy in responding to potential service disruptions. Thus, our

study aims to understand any differential impact of Southwest Airlines compared to other LCCs.

To do this, we define a categorical variable, 𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 𝑗 , with three distinct categories as follows:

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 𝑗 =

2,

1,

0,






if the airline is a full-service carrier

if the airlines is Southwest Airlines

otherwise

7The International Air Transport Association (IATA) designator codes are provided within parentheses.

138

Following prior studies in airline research (Nicolae et al., 2017; Deshpande and Arıkan, 2012;

Ramdas and Williams, 2006; Mayer and Sinai, 2003; Mazzeo, 2003), we account for fixed effects

by including dummy variables for route, year, month, day of the week, carrier, and departure and

arrival time blocks. First, the variable 𝑅𝑜𝑢𝑡𝑒𝑘 captures fixed effects for all origin–destination

airport pair combinations. The variable 𝐷𝑎𝑦_𝑜 𝑓 _𝑊 𝑒𝑒𝑘𝑡 indicates the day of the week of the

flight, accounting for fluctuations in flight volume throughout the week. 𝐷𝑒 𝑝_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘𝑡 and

𝐴𝑟𝑟_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘𝑡 represent one-hour time intervals based on the scheduled departure and arrival

times, respectively, to control for time-of-day effects. Finally, the variables 𝑌 𝑒𝑎𝑟𝑡 and 𝑀𝑜𝑛𝑡ℎ𝑡

control for annual and seasonal variations, respectively. In the mishandled baggage analysis, the

dataset is at the monthly level and the variables 𝑌 𝑒𝑎𝑟𝑦 and 𝑀𝑜𝑛𝑡ℎ𝑚 are included to account for

annual and seasonal variations, respectively. Table 4.2 presents a comprehensive overview of all

variables used in this study, and Table 4.3 reports the associated descriptive statistics.

Table 4.2 Variable Descriptions

Variable

Description

Database

Dependent Variables

𝐷𝑒 𝑝_𝐷𝑒𝑙𝑎𝑦𝑖𝑡 𝑘

The difference in minutes between a flight’s scheduled

OTP Dataset

and actual departure times.

𝐴𝑟𝑟_𝐷𝑒𝑙𝑎𝑦𝑖𝑡 𝑘

The difference in minutes between a flight’s scheduled

OTP Dataset

and actual arrival times.

𝑂𝑇 𝑃𝑖𝑡 𝑘

A dummy variable coded as 1 if a flight arrives within 15

OTP Dataset

minutes of its scheduled arrival time, and 0 otherwise.

𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛𝑖𝑡 𝑘

A dummy variable coded as 1 if a flight is cancelled, and

OTP Dataset

0 otherwise.

𝐵𝑎𝑔_𝐼𝑠𝑠𝑢𝑒 𝑗 𝑦𝑚

The number of mishandled bags per 1,000 bags enplaned. Air Travel Consumer Report

Explanatory Variables

𝐹𝑆𝐶 𝑗

A dummy variable coded as 1 for FSCs and 0 for LCCs.

OTP Dataset

139

Table(cid:1)4.2(cid:1)(cont’d)

Variable

Description

Database

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 𝑗

A categorical variable coded as 2 for FSCs, 1 for South-

OTP Dataset

west Airlines, and 0 for LCCs.

𝑅𝑜𝑢𝑡𝑒𝑘

An origin-destination airport pair combination.

𝐷𝑎𝑦_𝑜 𝑓 _𝑊 𝑒𝑒𝑘𝑡

The day of week of the flight.

OTP Dataset

OTP Dataset

𝐷𝑒 𝑝_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘𝑡 One-hour time block based on the scheduled departure

OTP Dataset

time of the flight.

𝐴𝑟𝑟_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘𝑡

One-hour time block based on the scheduled arrival time

OTP Dataset

𝑌 𝑒𝑎𝑟𝑡

𝑀𝑜𝑛𝑡ℎ𝑡

𝑌 𝑒𝑎𝑟 𝑦

𝑀𝑜𝑛𝑡ℎ𝑚

of the flight.

The year of the flight.

The month of the flight.

OTP Dataset

OTP Dataset

The year in which the air carrier operated.

Air Travel Consumer Report

The month in which the air carrier operated.

Air Travel Consumer Report

Table 4.3 Summary Statistics

Variable

Mean Std. dev. Min Max No. of obs.

Dependent Variable:
𝐷𝑒 𝑝_𝐷𝑒𝑙𝑎𝑦
𝐴𝑟𝑟_𝐷𝑒𝑙𝑎𝑦
𝑂𝑇 𝑃
𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛
𝐵𝑎𝑔_𝐼𝑠𝑠𝑢𝑒

8.855
3.899
0.818
0.014
5.451

1930
37.889
1910
40.681
1
0.386
0.119
1
1.628 3.240 11.300

-48
-107
0
0

3,023,984
3,023,984
3,023,984
3,104,887
98

Notes: Due to the number of observations available for flights with departure and arrival
delays as well as OTP, our analysis focuses on flight records that include these performance
metrics. As a result, gaps arise between this subset and the data that include cancelled flights.

4.5 Empirical Specification & Results

This section begins by outlining the sample composition used in the empirical analysis. We then

describe the matching procedure used to identify and select comparable routes operated by multiple

airlines with structurally different network models. Finally, we present the empirical models for

hypothesis testing, which form the foundation of our analysis.

140

4.5.1 Sample

Consistent with prior studies (Manchiraju et al., 2023; Deshpande and Arıkan, 2012; Ramdas

and Williams, 2006), we utilize individual flight-level data to test our hypotheses related to departure

delays, arrival delays, OTP, and cancellations. For the period from January 1, 2016, to December

31, 2019, we draw on the OTP dataset, which contains detailed flight-level information for seven

major U.S.-based carriers – AA, DL, UA, WN, NK, B6, and F9 – representing both full-service

and low-cost carrier types. In contrast, our analysis of mishandled baggage is conducted at the

carrier-month level due to the unavailability of more granular data. As discussed earlier, because of

inconsistencies in the baggage-handling metric reported by the U.S. DOT, we restrict this analysis

to the period from January 2019 to February 2020, during which the reporting metric is consistent

and reliable.

4.5.2 Matching

Airlines typically operate distinct route networks, which can pose challenges for direct perfor-

mance comparisons. For example, the Atlanta (ATL) – Orlando (MCO) route is predominantly

served by Delta Air Lines, whereas JetBlue Airways does not operate on this route. Similarly,

United Airlines frequently operates the Denver (DEN) – Minneapolis/St. Paul (MSP) route, while

Spirit Airlines does not offer service along this route. Such network asymmetries introduce route-

level heterogeneity, making direct comparisons of operational performance potentially biased. To

address this issue, we implement a route-matching procedure designed to control for unobserved

heterogeneity by focusing on routes that are either commonly served or closely comparable across

carriers. This approach enables more valid comparisons of performance outcomes by aligning the

analysis around operationally similar or equivalent routes.

The matching process proceeds as follows. First, we identify the top 10% of routes operated by

the four LCCs based on flight frequency over the 2016–2019 period. These high-frequency LCC

routes serve as the basis for comparative analysis. Next, we search for exact route matches between

these LCC routes and those operated by FSCs. When exact matches are unavailable, we extend

the matching criteria to include routes that connect geographically proximate airports within the

141

same metropolitan area. For instance, JetBlue operates flights between Boston Logan International

Airport (BOS) and Ronald Reagan Washington National Airport (DCA), while United operates

between BOS and Washington Dulles International Airport (IAD). Although DCA and IAD differ

in location and airport characteristics, both serve the Washington, D.C. metropolitan area and offer

comparable route lengths and demand patterns. Thus, BOS–DCA and BOS–IAD are considered

equivalent for the purpose of our analysis.

We further expand the matching framework to incorporate geographically aligned one-stop

connecting routes operated by FSCs, corresponding to nonstop LCC routes. For example, while

JetBlue offers direct service between John F. Kennedy International Airport (JFK) and Orlando

International Airport (MCO), American Airlines provides a comparable one-stop service via Char-

lotte Douglas International Airport (CLT), flying JFK–CLT–MCO. Such connecting itineraries are

treated as equivalent matches when they approximate the origin-destination pair served by LCCs. In

summary, our final analysis of OTP data includes only those routes identified through the extended

matching procedure, including exact matches, metropolitan-equivalent routes, and comparable one-

stop connecting itineraries. This refined approach enables consistent and meaningful comparisons

across the seven major carriers included in our study during the 2016–2019 period.

4.5.3 Empirical Models

To test H1A, H1B, H1C, and H1D, we employ the following ordinary least squares (OLS) linear

regression models, as specified in Equations (4.1) and (4.2).

Dep_Delay𝑖𝑡𝑘 = 𝛼0 + 𝛼1FSC𝑖𝑡𝑘 + 𝛼2Route𝑘 + 𝛼3Day_Of _Week𝑡 + 𝛼4Dep_Time_Block𝑡

+ 𝛼5Arr_Time_Block𝑡 + 𝛼6Year𝑡 + 𝛼7Month𝑡 + 𝜖𝑖𝑡𝑘

Arr_Delay𝑖𝑡𝑘 = 𝛽0 + 𝛽1FSC𝑖𝑡𝑘 + 𝛽2Route𝑘 + 𝛽3Day_Of _Week𝑡 + 𝛽4Dep_Time_Block𝑡

+ 𝛽5Arr_Time_Block𝑡 + 𝛽6Year𝑡 + 𝛽7Month𝑡 + 𝜔𝑖𝑡𝑘

(4.1)

(4.2)

The coefficients 𝛼1 and 𝛽1 in Equations (4.1) and (4.2) capture the differences in departure and

arrival delays between FSCs and LCCs, respectively.

142

We test H2A, H2B, H3A, and H3B using logistic regression models, as outlined in Equa-

tions (4.3) and (4.4). We use 𝑂𝑇 𝑃𝑖𝑡𝑘 and 𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛𝑖𝑡𝑘 as the dependent variables for testing

hypotheses H2A and H2B, and H3A and H3B, respectively.

log

(cid:18) Pr(OTP𝑖𝑡𝑘 = 1)
Pr(OTP𝑖𝑡𝑘 = 0)

(cid:19)

= 𝛿0 + 𝛿1FSC𝑖𝑡𝑘 + 𝛿2Route𝑘 + 𝛿3Day_Of _Week𝑡

+ 𝛿4Dep_Time_Block𝑡 + 𝛿5Arr_Time_Block𝑡 + 𝛿6Year𝑡

+ 𝛿7Month𝑡 + 𝜏𝑖𝑡𝑘

log

(cid:18) Pr(Cancellation𝑖𝑡𝑘 = 1)
Pr(Cancellation𝑖𝑡𝑘 = 0)

(cid:19)

= 𝜃0 + 𝜃1FSC𝑖𝑡𝑘 + 𝜃2Route𝑘 + 𝜃3Day_Of _Week𝑡

+ 𝜃4Dep_Time_Block𝑡 + 𝜃5Arr_Time_Block𝑡 + 𝜃6Year𝑡

+ 𝜃7Month𝑡 + 𝜂𝑖𝑡𝑘

(4.3)

(4.4)

In Equations (4.3) and (4.4), the parameters 𝛿1 and 𝜃1 estimate the differences in the odds of on-time

arrivals and cancellations between FSCs and LCCs, respectively.

Lastly, we use Equation (4.5) to test Hypothesis H4A and H4B.

Bag_Issue 𝑗 𝑦𝑚 = 𝛾0 + 𝛾1FSC 𝑗 𝑦𝑚 + 𝛾2Year𝑦 + 𝛾3Month𝑚 + 𝜁 𝑗 𝑦𝑚

(4.5)

In the above equation, the parameter 𝛾1 identifies the difference in the baggage handling issues

between FSCs and LCCs.

4.5.4 Results

In this section, we discuss the results of all our hypothesis tests. We provide the results for all

models, both with and without fixed effects.

As shown in Table 4.4, there are substantial differences in the magnitude of both departure and

arrival delays between FSCs and LCCs. To be specific, the estimated coefficients of 𝐹𝑆𝐶𝑖𝑡𝑘 in Model

1b (𝛼1 = −4.573 minutes, 𝑝 < 0.001) and Model 2b (𝛽1 = −5.354 minutes, 𝑝 < 0.001) suggest that

FSCs are significantly more effective than LCCs in managing flight delays. Accordingly, the findings

support H1A and H1C, and they demonstrate that FSCs consistently outperform LCCs in mitigating

143

both departure and arrival delays. Taken together, these findings indicate that FSCs enhance

operational performance by strategically reallocating internal resources in response to disruptions,

thereby enabling more rapid restoration of service quality during operational breakdowns. Given

the additioanl availability of resources, such as aircraft and maintenance personnel, administrative

and management staff, and specialized tools and equipment, FSCs are generally better equipped than

their LCC counterparts to manage flight delays resulting from service failures in airline operations.

Table 4.4 OLS Regression Results: Departure and Arrival Delays

Dependent Variable:

𝐷𝑒 𝑝_𝐷𝑒𝑙𝑎𝑦

𝐴𝑟𝑟_𝐷𝑒𝑙𝑎𝑦

𝐹𝑆𝐶

𝑅𝑜𝑢𝑡𝑒
𝐷𝑎𝑦_𝑜 𝑓 _𝑊 𝑒𝑒𝑘
𝐷𝑒 𝑝_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝐴𝑟𝑟_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝑌 𝑒𝑎𝑟
𝑀𝑜𝑛𝑡ℎ

No. of airlines
No. of obs.
Adj. 𝑅2

Model 1a Model 1b Model 2a Model 2b

-2.491***
(0.045)
No
No
No
No
No
No

7
3,023,984
0.001

-4.573***
(0.079)
Yes
Yes
Yes
Yes
Yes
Yes

7
3,023,984
0.037

-2.930*** -5.354***

(0.049)
No
No
No
No
No
No

(0.084)
Yes
Yes
Yes
Yes
Yes
Yes

7
3,023,984
0.001

7
3,023,984
0.036

Notes: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical
significance at the 10 percent, 5 percent, 1 percent, and 0.1 percent levels. Robust
standard errors are reported in parentheses below parameter estimates.

The results for the hypothesis tests of OTP and cancellations are similar. As shown in Table 4.5,

the coefficient of 𝐹𝑆𝐶𝑖𝑡𝑘 in Model 3b of Table 4.5 (𝛿1 = 0.431, 𝑝 < 0.001) is statistically

significant at the 0.1% level. Consistent with the findings on departure and arrival delays, this

result indicates that FSCs are more effective than LCCs in maintaining OTP, and therefore more

capable of controlling longer delays. These findings provide empirical support for hypothesis

H2A. To facilitate interpretation, we discuss the result in terms of log-odds, which appear on the

left-hand side of Equation (4.3). In logistic regression, the log-odds of the event occurring, defined

as log

(cid:17)

(cid:16) 𝑝
1−𝑝

, where 𝑝 = Pr(𝑂𝑇 𝑃𝑖𝑡𝑘 = 1), increases by the value of the estimated coefficient.

Accordingly, transitioning from an LCC (𝐹𝑆𝐶𝑖𝑡𝑘 = 0) to an FSC (𝐹𝑆𝐶𝑖𝑡𝑘 = 1) results in a 0.431

144

increase in the log-odds of a flight arriving on time. To interpret this effect in terms of odds, we

exponentiate the coefficient: 𝑒0.431 = 1.539. This implies that the odds of a flight arriving within

15 minutes of its scheduled arrival time are 53.9% higher for FSCs compared to LCCs.

Similarly, the coefficient of 𝐹𝑆𝐶𝑖𝑡𝑘 in Model 4b of Table 4.5 (𝜃1 = −0.462, 𝑝 < 0.001)

indicates that FSCs are significantly less likely to cancel scheduled flights compared to LCCs,

further reinforcing their operational reliability. This finding provides empirical support for H3A.

As with 𝑂𝑇 𝑃𝑖𝑡𝑘 , we interpret this result in terms of log-odds, as shown on the left-hand side of

Equation (4.4). Moving from an LCC (𝐹𝑆𝐶𝑖𝑡𝑘 = 0) to an FSC (𝐹𝑆𝐶𝑖𝑡𝑘 = 1) decreases the log-odds

of flight cancellation by 0.462. Exponentiating this coefficient (𝑒−0.462 = 0.630) implies that the

odds of flight cancellation are 37.0% lower for FSCs relative to LCCs.

Table 4.5 Logistic Regression Results: OTP and Cancellations

Dependent Variable:

𝑂𝑇 𝑃

𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛

Model 3a Model 3b Model 4a Model 4b

𝐹𝑆𝐶

𝑅𝑜𝑢𝑡𝑒
𝐷𝑎𝑦_𝑜 𝑓 _𝑊 𝑒𝑒𝑘
𝐷𝑒 𝑝_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝐴𝑟𝑟_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝑌 𝑒𝑎𝑟
𝑀𝑜𝑛𝑡ℎ

0.241***
(0.003)
No
No
No
No
No
No

0.431***
(0.005)
Yes
Yes
Yes
Yes
Yes
Yes

No. of airlines
No. of obs.
McFadden’s Pseudo 𝑅2

7
3,023,984
0.002

7
3,023,984
0.056

-0.651***
(0.010)
No
No
No
No
No
No

7
3,104,887
0.009

-0.462***
(0.016)
Yes
Yes
Yes
Yes
Yes
Yes

7
3,104,887
0.061

Notes: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical
significance at the 10 percent, 5 percent, 1 percent, and 0.1 percent levels. Robust
standard errors are reported in parentheses below parameter estimates.

Finally, we present the results related to our baggage handling analysis. As shown in Table 4.6,

the estimated coefficient of 𝐹𝑆𝐶 𝑗 𝑦𝑚 in OLS regression Model 5b (𝛾1 = 1.954, 𝑝 < 0.001) suggests

that LCCs handle baggage more effectively than FSCs, with approximately two fewer mishandled

bags per 1,000 enplaned bags. These results provide empirical support for H4B. As discussed

in Section 4.3.4, FSCs often operate through major hub airports where congestion increases the

145

risk of service failures, particularly in baggage handling during connections. These inefficiencies

are exacerbated by complex scheduling, frequent transfers, and greater fleet diversity (Alan and

Lapré, 2018). In contrast, LCCs typically operate point-to-point networks with fewer connections,

reducing baggage transfers. This streamlined process lowers the likelihood of baggage mishandling,

as supported by our empirical findings.

Table 4.6 OLS Regression Results: Baggage Issues

Dependent Variable:

𝐵𝑎𝑔_𝐼𝑠𝑠𝑢𝑒

𝐹𝑆𝐶

𝑌 𝑒𝑎𝑟
𝑀𝑜𝑛𝑡ℎ

No. of airlines
No. of obs.
Adj. 𝑅2

Model 5a Model 5b

1.954***
(0.296)
No
No

1.954***
(0.284)
Yes
Yes

7
98
0.350

7
98
0.401

Notes: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and
∗ ∗ ∗𝑝 < 0.001 indicate statistical significance at
the 10 percent, 5 percent, 1 percent, and 0.1 per-
cent levels. Robust standard errors are reported in
parentheses below parameter estimates.

4.6 Additional Analysis

This section presents a distinct analysis of Southwest Airlines, considering it separately from

the other LCCs.

4.6.1 Empirical Specification

As discussed in Section 4.4.2.2, given the scale of its operations, stronger operational control

at select airports within its network, unique market positioning, distinctive implementation of its

baggage policy, and notable influence on industry and market competition, analyzing Southwest

Airlines separately allows for more nuanced insights into how its performance compares with both

FSCs and other LCCs. To conduct this analysis, we use the same set of dependent variables

introduced in Section 4.5.3 and apply the corresponding empirical specifications presented in

Equations (4.1) through (4.5). However, instead of using 𝐹𝑆𝐶𝑖𝑡𝑘 as the primary independent

146

variable of interest, we employ an alternative categorical variable, 𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦𝑖𝑡𝑘 , as defined in

Section 4.4.2.2.

4.6.2 Results

As shown in Table 4.7, there are significant differences in both departure and arrival delay

performance across carrier types. As expected, FSCs exhibit significantly lower departure and

arrival delays compared to the other groups. The estimates from the model without fixed effects

(Model 6a) indicate that Southwest Airlines has lower departure delays than other LCCs, but

higher delays than FSCs. However, in the fixed-effects specification (Model 6b), the difference in

departure delays between Southwest and other LCCs is no longer statistically significant. Further

analysis suggests that this result is largely driven by the inclusion of route fixed effects, a finding

that warrants additional investigation. Turning to arrival delays, the results from Models 7a and 7b

show that Southwest performs significantly better than other LCCs, albeit worse than FSCs. One

possible explanation is that Southwest’s stronger operational presence and control at many of the

airports in its network may enhance its ability to manage arrival delays more effectively.

The results from Models 8a and 8b presented in Table 4.8 indicate that Southwest Airlines

performs worse than FSCs and is statistically indistinguishable from other LCCs in terms of OTP.

Interestingly, the results from Models 9a and 9b reveal that Southwest performs worse not only

than FSCs but also than other LCCs in managing flight cancellations. Based on the parameter

estimates, a shift from another LCC to Southwest is associated with a 0.793 increase in the log-

odds of cancellation. This indicates that the odds of cancellation are 121.0% higher for Southwest

compared to other LCCs.

Southwest’s comparatively higher cancellation rate among LCCs may stem from its operational

scale and network complexity, which exceed those of most peer carriers. Unlike other LCCs that

operate fewer flights and maintain simpler schedules, Southwest’s dense point-to-point network

leaves it more vulnerable to cascading disruptions. Additionally, its tight turnaround times limit

flexibility during irregular operations. Southwest may also adopt a more proactive cancellation

strategy, opting to cancel flights earlier to preserve schedule integrity, which can increase the

147

Table 4.7 Categorical Regression Results: Departure and Arrival Delays

Dependent Variable:

𝐷𝑒 𝑝_𝐷𝑒𝑙𝑎𝑦

𝐴𝑟𝑟_𝐷𝑒𝑙𝑎𝑦

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 = 𝐹𝑆𝐶

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 = 𝑆𝑜𝑢𝑡ℎ𝑤𝑒𝑠𝑡

𝑅𝑜𝑢𝑡𝑒
𝐷𝑎𝑦_𝑜 𝑓 _𝑊 𝑒𝑒𝑘
𝐷𝑒 𝑝_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝐴𝑟𝑟_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝑌 𝑒𝑎𝑟
𝑀𝑜𝑛𝑡ℎ

No. of airlines
No. of obs.
Adj. 𝑅2

Model 6a Model 6b Model 7a Model 7b

-5.436***
(0.086)
-4.615***
(0.088)
No
No
No
No
No
No

7
3,023,984
0.002

-4.577***
(0.102)
-0.011
(0.131)
Yes
Yes
Yes
Yes
Yes
Yes

7
3,023,984
0.037

-5.381***
(0.092)
-3.843***
(0.094)
No
No
No
No
No
No

7
3,023,984
0.002

-6.232***
(0.109)
-2.655***
(0.140)
Yes
Yes
Yes
Yes
Yes
Yes

7
3,023,984
0.036

Notes: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical
significance at the 10 percent, 5 percent, 1 percent, and 0.1 percent levels. An ordinary
least squares (OLS) linear regression model is used. Robust standard errors (s.e.) are
reported in parentheses below parameter estimates.

number of cancellations even if it benefits downstream recovery.

Finally, we analyze the number of mishandled bags per 1,000 enplaned bags (𝐵𝑎𝑔_𝐼𝑠𝑠𝑢𝑒 𝑗 𝑦𝑚).

As reported in Table 4.9, the estimate of 𝐹𝑆𝐶 in OLS regression Model 10b is 1.837 (𝑝 < 0.001),

indicating that LCCs manage baggage handling more effectively than FSCs. Moreover, the estimate

of 𝑆𝑜𝑢𝑡ℎ𝑤𝑒𝑠𝑡 is –0.470 (𝑝 < 0.01), suggesting that Southwest Airlines outperforms both FSCs and

other LCCs in minimizing baggage mishandling incidents.

Airlines that impose fees for checked baggage typically observe a decrease in the volume of

checked bags, which can mitigate baggage handling complexity and, in turn, reduce the likelihood

of departure delays. This suggests that fewer checked bags could benefit not only the operating

carrier but also other airlines operating at the same airport (Nicolae et al., 2017).

In contrast,

Southwest Airlines has promoted its well-known “Bags Fly Free” campaign, which has led to

a higher volume of checked baggage as a natural consequence. When evaluating the absolute

number of mishandled baggage incidents, Southwest may appear more vulnerable to baggage-

148

Table 4.8 Categorical Regression Results: OTP and Cancellations

Dependent Variable:

𝑂𝑇 𝑃

𝐶𝑎𝑛𝑐𝑒𝑙𝑙𝑎𝑡𝑖𝑜𝑛

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 = 𝐹𝑆𝐶

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 = 𝑆𝑜𝑢𝑡ℎ𝑤𝑒𝑠𝑡

𝑅𝑜𝑢𝑡𝑒
𝐷𝑎𝑦_𝑜 𝑓 _𝑊 𝑒𝑒𝑘
𝐷𝑒 𝑝_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝐴𝑟𝑟_𝑇𝑖𝑚𝑒_𝐵𝑙𝑜𝑐𝑘
𝑌 𝑒𝑎𝑟
𝑀𝑜𝑛𝑡ℎ

Model 8a Model 8b Model 9a Model 9b

0.407***
(0.005)
0.267***
(0.005)
No
No
No
No
No
No

0.427***
(0.006)
-0.014
(0.009)
Yes
Yes
Yes
Yes
Yes
Yes

-0.420*** -0.274***

(0.015)
0.342***
(0.016)
No
No
No
No
No
No

(0.019)
0.793***
(0.035)
Yes
Yes
Yes
Yes
Yes
Yes

No. of airlines
No. of obs.
McFadden’s Pseudo 𝑅2

7
3,023,984
0.003

7
3,023,984
0.056

7
3,104,887
0.011

7
3,104,887
0.063

Notes: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and ∗ ∗ ∗𝑝 < 0.001 indicate statistical
significance at the 10 percent, 5 percent, 1 percent, and 0.1 percent levels. A logistic
regression model is used. Robust standard errors (s.e.) are reported in parentheses
below parameter estimates.

Table 4.9 Categorical Regression Results: Baggage Issues

Dependent Variable:

𝐵𝑎𝑔_𝐼𝑠𝑠𝑢𝑒

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 = 𝐹𝑆𝐶

𝐶𝑎𝑡𝑒𝑔𝑜𝑟 𝑦 = 𝑆𝑜𝑢𝑡ℎ𝑤𝑒𝑠𝑡

𝑌 𝑒𝑎𝑟
𝑀𝑜𝑛𝑡ℎ

No. of airlines
No. of obs.
Adj. 𝑅2

Model 10a Model 10b

1.837***
(0.304)
-0.470*
(0.209)
No
No

7
98
0.352

1.837***
(0.294)
-0.470**
(0.161)
Yes
Yes

7
98
0.405

Notes: +𝑝 < 0.1, ∗𝑝 < 0.05, ∗ ∗ 𝑝 < 0.01, and
∗ ∗ ∗𝑝 < 0.001 indicate statistical significance at the 10
percent, 5 percent, 1 percent, and 0.1 percent levels. An
ordinary least squares (OLS) linear regression model is
used. Robust standard errors are reported in parentheses
below parameter estimates.

149

related disruptions, which could contribute to increased departure delays and lower OTP. However,

when mishandling is assessed in relative terms as the number of mishandled bags per 1,000 enplaned

bags, the interpretation shifts, as demonstrated in our analysis. Based on the empirical evidence

in Tables 4.7, 4.8, and 4.9, although Southwest Airlines does not exhibit a distinct advantage over

other LCCs in managing departure delays or OTP, and even performs worse in managing flight

cancellations, it significantly outperforms both FSCs and other LCCs in handling baggage issues

when measured on a proportional basis. This suggests that, despite the perceived operational

disadvantage introduced by its baggage fee policy, Southwest Airlines effectively manages baggage

handling performance as evidenced by our results. This suggests that Southwest has developed

highly efficient baggage handling processes that enable it to maintain superior performance even

under heavier operational loads.

4.7 Discussion

In the airline industry, carriers design flight routes to optimize operational efficiency. Network

strategies are typically shaped by key operational priorities such as cost efficiency, route flexibility,

service coverage, and revenue. The two predominant network structures are the point-to-point

and hub-and-spoke systems. In a point-to-point network, passengers travel directly between cities,

while in a hub-and-spoke system, they often connect through a hub city when neither the origin nor

destination is a hub. Most FSCs primarily operate under a hub-and-spoke model, whereas LCCs

tend to rely more heavily on point-to-point networks (Mellat-Parast et al., 2015; Cook and Goodwin,

2008). These strategic choices are further influenced by broader considerations, including network

configuration, product offerings, pricing models, and resource allocation decisions (Bourjade et al.,

2017).

In this study, our first research question investigates how airlines operating under different

network structures manage service quality. Prior research has not comprehensively examined

how KPIs reflecting operational performance under disruptions capture the effectiveness of airline

operations across different network structures. This study aims to demonstrate that service quality

outcomes are closely linked to airlines’ operational strategies, particularly as shaped by their

150

network configurations. Building on this, we also examine Southwest Airlines separately, as it

is widely recognized as one of the largest U.S. air carriers in terms of operational volume and

capacity, despite being formally classified as an LCC. To test our hypotheses, we use historical data

on operational and service performance for seven U.S. domestic airlines, drawing on OTP data and

mishandled baggage records obtained from the official websites of the BTS and the U.S. DOT.

First, our results reveal significant differences between FSCs and LCCs in managing both

departure and arrival delays, with FSCs consistently outperforming LCCs in minimizing overall

delays. Additionally, while Southwest Airlines performs comparably to other LCCs – and worse

than FSCs – in managing departure delays, it outperforms other LCCs in managing arrival delays.

This pattern suggests that Southwest’s stronger operational presence and control at several airports

within its network may contribute to its improved performance in mitigating arrival delays. We

also evaluate airline operational performance based on on-time arrival performance and flight

cancellations. The results indicate that FSCs consistently outperform LCCs in both maintaining

OTP and minimizing flight cancellations. Specifically, the odds of a flight arriving within 15

minutes of its scheduled arrival time are 53.9% higher for FSCs compared to LCCs and the odds

of a flight being canceled are 37% lower for FSCs. Moreover, we find no statistically significant

difference in the OTP of Southwest compared to other LCCs. However, Southwest performs

substantially worse than both FSCs and its LCC peers in managing flight cancellations.

These findings highlight the structural advantages of the hub-and-spoke network used by FSCs,

which includes built-in redundancy such as reserve crews, spare aircraft, and extensive rebooking

options, all of which lead to greater schedule reliability, higher OTP, and lower delays and cancella-

tions. In contrast, the leaner, point-to-point networks operated by LCCs, including Southwest, may

lack sufficient slack to absorb disruptions, particularly when irregular operations arise. Southwest’s

notably poor performance on cancellations, despite having comparable OTP to other LCCs, sug-

gests that its operational model, though efficient under normal conditions, may be more vulnerable

when recovery resources are constrained. This highlights a critical trade-off in network design:

while point-to-point structures can offer flexibility and efficiency, they may do so at the expense of

151

robustness under strain.

These results have practical implications for both airline strategy and policy. As disruptions

become more frequent and customer expectations for service reliability continue to rise, achieving

an effective balance between efficiency and operational resilience is increasingly critical for airlines.

For managers, investing in targeted buffers, operational contingencies, and selective redundancies,

such as reserve crews or aircraft and increased operational control at key airports, can enhance

operational performance without fundamentally altering the cost structure or compromising the

efficiency advantages of a point-to-point model.

Another important operational performance metric examined in this study is the mishandled

baggage rate, a key indicator of service failures in the airline industry. Mishandled baggage incidents

not only reflect lapses in service quality but also impose additional operational costs on airlines,

such as those related to locating lost items and arranging their timely return to passengers. These

failures, defined as instances where the delivered service falls short of customers’ expectations,

can significantly erode customer satisfaction and, in turn, impact long-term loyalty (Mellat-Parast

et al., 2015).

In this study, we investigate how airline network structure influences baggage-related service

failures. Our empirical findings reveal that LCCs are generally more effective than FSCs in handling

baggage, with LCCs reporting approximately two fewer mishandled bags per 1,000 enplaned bags.

This advantage may stem from the leaner operational model of LCCs, characterized by direct

routes, and fewer baggage transfers, which collectively reduce the likelihood of errors in baggage

handling. Southwest Airlines, in particular, outperforms both other LCCs and FSCs in minimizing

baggage-handling failures. A possible explanation is that, due to its no-fee checked baggage policy,

Southwest handles a disproportionately high volume of checked bags, and this operational exposure

may have led to the development of more efficient and experienced baggage handling processes.

These results offer important implications for airline operations. While FSCs offer higher

schedule reliability, LCCs achieve superior outcomes in specific quality dimensions, such as bag-

gage handling, through simplified logistics, higher process specialization, and operational focus.

152

For managers, these insights suggest that improving service reliability does not always require

extensive redundancy; rather, it may be achieved through targeted process optimization aligned

with the carrier’s operating model.

We contribute to the existing literature by systematically comparing the performance of the two

predominant airline network structures – hub-and-spoke versus point-to-point – across multiple

operational metrics, including departure delays, arrival delays, OTP, flight cancellations, and

mishandled baggage incidents. This multidimensional approach enables managers to evaluate

whether meaningful differences in performance arise from the underlying network design itself.

Furthermore, it provides a framework for benchmarking operational outcomes not only relative to

competitors employing different network strategies but also across the various service dimensions

within a firm’s own operations.

While this study offers valuable insights, it also presents several limitations that suggest promis-

ing directions for future research. First, we use the number of mishandled bags per 1,000 bags

enplaned as one of our service quality measures. Beginning in January 2019, the U.S. DOT revised

its reporting in the Air Travel Consumer Reports to reflect the number of bags enplaned, rather than

the number of enplaned passengers. Accordingly, we limit our analysis of mishandled baggage data

to the period from January 2019 to February 2020 to minimize the influence of disruptions result-

ing from the COVID-19 pandemic. As a result, the number of observations related to mishandled

baggage incidents is relatively small across the seven airlines included in this study. In summary,

future research would benefit from identifying ways to expand the sample size, particularly for

metrics related to mishandled baggage per enplaned bag, to strengthen the robustness of findings.

Furthermore, this study does not focus exclusively on periods of extreme disruption to examine

the influence of airlines’ network strategies on operational performance metrics. However, given

that the COVID-19 pandemic has been widely recognized as one of the most disruptive events in

the service industry, future research could yield deeper insights by analyzing performance under

both normal operating conditions and periods characterized by systemic disruption. Analyzing

performance across different levels of disruptions would enable a more comprehensive assessment

153

of how network structures influence operational outcomes under uncertainty. Finally, although our

models include a wide range of fixed effects related to airline operations and scheduling, aimed

at isolating the impact of the primary independent variables on service quality outcomes, it is

important to acknowledge additional relevant factors. Consistent with prior research on airline

operations (Nicolae et al., 2017; Prince and Simon, 2015; Ramdas et al., 2013), incorporating

control variables such as weather conditions and load factors could enhance the explanatory power

of the analysis, as these factors are closely linked to the overall success and reliability of airline

operations. The aforementioned limitations raise important research questions and warrant further

in-depth investigation in future studies.

154

REFERENCES

Alan, Y. and Lapré, M. A. (2018). Investigating operational predictors of future financial distress

in the US airline industry. Production and Operations Management, 27(4):734–755.

Alderighi, M. and Gaggero, A. A. (2018). Flight cancellations and airline alliances: Empirical
evidence from Europe. Transportation Research Part E: Logistics and Transportation Review,
116:90–101.

Arıkan, M., Deshpande, V., and Sohoni, M. (2013). Building reliable air-travel infrastructure using
empirical data and stochastic models of airline networks. Operations Research, 61(1):45–64.

Barla, P. and Constantatos, C. (2000). Airline network structure under demand uncertainty. Trans-

portation Research Part E: Logistics and Transportation Review, 36(3):173–180.

Barla, P. and Constantatos, C. (2005). Strategic interactions and airline network morphology under

demand uncertainty. European Economic Review, 49(3):703–716.

Barnhart, C., Fearing, D., Odoni, A., and Vaze, V. (2012). Demand and capacity management in

air transportation. EURO Journal on Transportation and Logistics, 1(1-2):135–155.

Bauranov, A., Parks, S., Jiang, X., Rakas, J., and González, M. C. (2021). Quantifying the
resilience of the us domestic aviation network during the COVID-19 pandemic. Frontiers in
Built Environment, 7:642295.

Bitzan, J. and Peoples, J. (2016). A comparative analysis of cost change for low-cost, full-service,
and other carriers in the US airline industry. Research in Transportation Economics, 56:25–41.

Bourjade, S., Huc, R., and Muller-Vibes, C. (2017). Leasing and profitability: Empirical evidence

from the airline industry. Transportation Research Part A: Policy and Practice, 97:30–46.

Brueckner, J. K. (2004). Network structure and airline scheduling. The Journal of Industrial

Economics, 52(2):291–312.

Burton, R. M. and Obel, B. (2018). The science of organizational design: Fit between structure

and coordination. Journal of Organization Design, 7(1):1–13.

Campbell, J. P., Bownas, D. A., Peterson, N. G., and Dunnette, M. D. (1974). The measurement
of organizational effectiveness: A review of relevant research and opinion. Navy Personnel
Research and Development Center.

Cao, K. H., Krier, B., Liu, C.-M., McNamara, B., and Sharpe, J. (2017). The nonlinear effects of
market structure on service quality: Evidence from the US airline industry. Review of Industrial
Organization, 51:43–73.

155

Chang, M.-H. and Harrington, J. E. (2000). Centralization vs. decentralization in a multi-unit orga-
nization: A computational model of a retail chain as a multi-agent adaptive system. Management
Science, 46(11):1427–1440.

Ciliberto, F., Cook, E. E., and Williams, J. W. (2019). Network structure and consolidation in the

US airline industry, 1990–2015. Review of Industrial Organization, 54:3–36.

Cook, G. N. and Goodwin, J. (2008). Airline networks: A comparison of hub-and-spoke and
point-to-point systems. Journal of Aviation/Aerospace Education & Research, 17(2):51–60.

Dalton, D. R., Todor, W. D., Spendolini, M. J., Fielding, G. J., and Porter, L. W. (1980). Organization
structure and performance: A critical review. Academy of Management Review, 5(1):49–64.

Deshpande, V. and Arıkan, M. (2012). The impact of airline flight schedules on flight delays.

Manufacturing & Service Operations Management, 14(3):423–440.

Ding, Y., Wandelt, S., Wu, G., Xu, Y., and Sun, X. (2023). Towards efficient airline disruption recov-
ery with reinforcement learning. Transportation Research Part E: Logistics and Transportation
Review, 179:103295.

Girod, S. J. and Whittington, R. (2017). Reconfiguration, restructuring and firm performance:
Dynamic capabilities and environmental dynamism. Strategic Management Journal, 38(5):1121–
1133.

Gualini, A., Zou, L., and Dresner, M. (2023). Airline strategies during the pandemic: What

worked? Transportation Research Part A: Policy and Practice, 170:103625.

Hassan, L., Santos, B. F., and Vink, J. (2021). Airline disruption management: A literature review

and practical challenges. Computers & Operations Research, 127:105137.

James, L. R. and Jones, A. P. (1976). Organizational structure: A review of structural dimen-
sions and their conceptual relationships with individual attitudes and behavior. Organizational
Behavior and Human Performance, 16(1):74–113.

Kohl, N., Larsen, A., Larsen, J., Ross, A., and Tiourine, S. (2007). Airline disruption manage-
ment—perspectives, experiences and outlook. Journal of Air Transport Management, 13(3):149–
162.

Lordan, O. (2014). Study of the full-service and low-cost carriers network configuration. Journal

of Industrial Engineering and Management (JIEM), 7(5):1112–1123.

Malladi, V. and Sohoni, M. (2022). Data-driven modeling of delay propagation in air transportation

networks. Working Paper.

Manchiraju, C., Sohoni, M. G., and Deshpande, V. (2023).

It’s not simply luck: The impact

156

of network strategy, schedule padding, and operational improvements on domestic on-time
performance in the US airline industry. Production and Operations Management, 32(11):3559–
3574.

Mayer, C. and Sinai, T. (2003). Network effects, congestion externalities, and air traffic delays: Or

why not all delays are evil. American Economic Review, 93(4):1194–1215.

Mazzeo, M. J. (2003). Competition and service quality in the US airline industry. Review of

industrial Organization, 22:275–296.

Mellat-Parast, M., Golmohammadi, D., McFadden, K. L., and Miller, J. W. (2015). Linking
business strategy to service failures and financial performance: Empirical evidence from the US
domestic airline industry. Journal of Operations Management, 38:14–24.

Miller, D. (1992). Environmental fit versus internal fit. Organization Science, 3(2):159–178.

Nicolae, M., Arıkan, M., Deshpande, V., and Ferguson, M. (2017). Do bags fly free? An
empirical analysis of the operational implications of airline baggage fees. Management Science,
63(10):3187–3206.

Oum, T. H., Zhang, A., and Zhang, Y. (1995). Airline network rivalry. Canadian Journal of

Economics, 28(4a):836–857.

Parast, M. and Golmohammadi, D. (2020). The impact of firm size and business strategy on response
to service disruptions: Evidence from the US domestic airline industry. IEEE Transactions on
Engineering Management, 69(5):1944–1957.

Park, H., Bellamy, M. A., and Basole, R. C. (2018). Structural anatomy and evolution of supply
chain alliance networks: A multi-method approach. Journal of Operations Management, 63:79–
96.

PeCoy, M. D. and Redmond, M. A. (2023). Flight reliability during periods of high uncertainty.

Journal of Air Transport Management, 106:102322.

Phillips, G. and Sertsios, G. (2013). How do firm financial conditions affect product quality and

pricing? Management Science, 59(8):1764–1782.

Prince, J. T. and Simon, D. H. (2009). Multimarket contact and service quality: Evidence from on-
time performance in the US airline industry. Academy of Management Journal, 52(2):336–354.

Prince, J. T. and Simon, D. H. (2015). Do incumbents improve service quality in response to entry?

Evidence from airlines’ on-time performance. Management Science, 61(2):372–390.

Ramdas, K. and Williams, J. (2006). An empirical investigation into the tradeoffs that impact

on-time performance in the airline industry. Washington Post, pages 1–32.

157

Ramdas, K., Williams, J., and Lipson, M. (2013). Can financial markets inform operational
improvement efforts? Evidence from the airline industry. Manufacturing & Service Operations
Management, 15(3):405–422.

Redmond, M., Campbell, A. M., and Ehmke, J. F. (2019). The most reliable flight itinerary problem.

Networks, 73(3):325–343.

Rupp, N., Owens, D., Plumly, L., et al. (2006). Does competition influence airline on-time

performance. Advances in Airline Economics, 1:251–272.

Rupp, N. G. and Holmes, G. M. (2006). An investigation into the determinants of flight cancella-

tions. Economica, 73(292):749–783.

Sheffi, Y. and Rice Jr, J. B. (2005). A supply chain view of the resilient enterprise. MIT Sloan

Management Review, 47(1):41–48.

Sodhi, M. S. and Chopra, S. (2004). Managing risk to avoid supply-chain breakdown. MIT Sloan

Management Review, 46(1):53–61.

Sugishita, K., Mizutani, H., and Hanaoka, S. (2024). Disruption and recovery of the us domes-
tic airline networks during the COVID-19 pandemic. Journal of Air Transport Management,
114:102504.

Talluri, S., Kull, T. J., Yildiz, H., and Yoon, J. (2013). Assessing the efficiency of risk mitigation

strategies in supply chains. Journal of Business Logistics, 34(4):253–269.

Tsikriktsis, N. (2007). The effect of operational performance and focus on profitability: A lon-
gitudinal study of the US airline industry. Manufacturing & Service Operations Management,
9(4):506–517.

Van de Ven, A. H. (1976). A framework for organization assessment. Academy of Management

Review, 1(1):64–78.

Zgodavová, Z., Rozenberg, R., and Szabo, S. (2018). Analysis of point-to-point versus hub-and-
spoke airline networks. In 2018 XIII International Scientific Conference-New Trends in Aviation
Development (NTAD), pages 158–163. IEEE.

Zou, L. and Yu, C. (2020). The evolving market entry strategy: A comparative study of Southwest

and JetBlue. Transportation Research Part A: Policy and Practice, 132:682–695.

158