INTERNATIONAL

TRANSACTIONS
IN OPERATIONAL
Intl. Trans. in Op. Res. 31 (2024) 1093–1121 RESEARCH
DOI: 10.1111/itor.13186

Operation strategy in an E-commerce platform supply chain:

whether and how to introduce live streaming services?
Xuemei Zhanga,b , Haoran Chena and Zhi Liuc,∗
a
School of Business, Fuyang Normal University, Fuyang, Anhui, China
b
Anhui Provincial Key Laboratory of Regional Logistics Planning and Modern Logistics Engineering,
Fuyang, Anhui, China
c
College of Management Engineering, Anhui Polytechnic University, Wuhu, Anhui, China
E-mail: [email protected] [Zhang]; [email protected] [Chen]; [email protected] [Liu]

Received 7 December 2021; received in revised form 3 July 2022; accepted 13 July 2022

Abstract
E-commerce platforms provide sales channels for enterprises to exploit more markets. Meanwhile,
E-commerce platform supply chain members are increasingly using live streaming services to boost prof-
its. However, whether to introduce live streaming services to E-commerce platform supply chain and which
operation strategy to implement is unclear. In this study, we construct three models to examine the oper-
ation strategy in an E-commerce platform supply chain by considering live streaming services. The results
show that introducing live streaming services is always beneficial for both the E-commerce platform and live
streaming services supplier. When live streaming services is introduced and an appropriate revenue sharing
proportion and signing fees are set by the E-commerce platform supply chain members, the operation strat-
egy of live streaming services introduced and implemented by the manufacturer is the consistent choice for
E-commerce platform supply chain members, which is also beneficial for E-commerce platform supply chain
system and consumers.

Keywords: supply chain management; E-commerce platform; live streaming services; operation strategy

1. Introduction

The rise of Internet information and mobile technology over the past few years led to the rapid
development of E-commerce. Many E-commerce platforms, such as Amazon, eBay, and JD, have
entered to the E-commerce market (Wang et al., 2019; Yoo and Jang, 2019; Shi et al., 2022). These
platforms and their suppliers make up E-commerce platform supply chains (EPSCs) (Tsunoda
and Zennyo, 2021; Guan et al., 2022). In EPSCs, E-commerce platforms act as retailers (reselling

∗
Corresponding author.

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA02148,
USA.
1094 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
mode), or provide a marketplace for manufacturers and charge a certain revenue sharing propor-
tion (platform selling mode) (Liu et al., 2020; Wang et al., 2022; Xu et al., 2022; Zhang et al.,
2022b). For example, Amazon announced revenues of $210 billion from reselling and $390 billion
from third-party sales in 2021.1 Platform selling mode has been widely used by more and more
enterprises.
Under the platform selling mode, the manufacturer could get in touch with consumers more
directly (Liu et al., 2021). However, consumer demands would be negatively influenced by their lack
of experience (Zhang et al., 2022a), while live streaming is a social media outlet in which streamers
can explain and sell products and services in real time (Kang et al., 2021). The application of live
streaming attracts more consumers and increases the vitality of the social economy. To increase
consumer engagement, many enterprises began to adopt live streaming as a sales channel (Guo
et al., 2021a). For example, a report published in 2022 showed that the number of live streaming
users in China reached 635 million in 2021.2 Moreover, according to a public report, the market of
live streaming commerce in China achieved ¥1.2 trillion in 2021, which is expected to rise to ¥1.51
trillion in 2022.3
The rapid growth of live streaming has a profound impact on social commerce (Singh et al.,
2021) and encourages enterprises in EPSCs to introduce live streaming services (LSSs) from LSSs
supplier. In live streaming shopping, LSSs enable consumers to obtain more detailed informa-
tion about products through real-time videos and allows sellers to answer consumer questions
(Wongkitrungrueng and Assarut, 2020). LSSs would not only increase the consumer demands
of live streaming channels, but also have a positive impact on E-commerce platforms, which
is called service spillover effect (Chai et al., 2020). For example, to increase the consumer de-
mands of live streaming and E-commerce platform channels, Amazon launched Amazon Live
Creator for third-party sellers to display product information and interact with consumers in real
time.4
In the operations of EPSCs, manufacturers or E-commerce platforms can introduce and adopt
the LSSs from LSSs suppliers. For example, Gree, a manufacturer in China, earned ¥47.6 bil-
lion from the live streaming channel in 2020 and ¥1.14 billion in 2021.5 E-commerce platforms
such as Taobao and JD are all deploying live streaming services.6 In practice, manufacturers or E-
commerce platforms would introduce and adopt the LSSs from LSSs suppliers by paying a certain
proportion of the revenue and signing fee, which forms a horizontal cooperation relationship,7,8

1
Available at https://baijiahao.baidu.com/s?id=1724818485657722150&wfr=spider&for=pc (Accessed 1 July 2022).
2
Available at https://www.iimedia.cn/c400/83735.html (Accessed 1 July 2022).
3
Available at https://www.askci.com/news/chanye/20220318/1451401746320_2.shtml (Accessed 1 July 2022).
4
Available at https://www.cnbc.com/2021/05/03/retailers-from-bloomingdales-to-petco-test-livestreaming-to-win-
sales.html (Accessed 1 July 2022).
5
Available at http://www.southmoney.com/caijing/caijingyaowen/202111/19125546.html (Accessed 1 July 2022).
6
Available at https://www.300.cn/dspd/341064.html (Accessed 1 July 2022).
7
Available at https://zhuanlan.zhihu.com/p/262053824 (Accessed 1 July 2022).
8
Available at https://www.sohu.com/a/133710602_633472 (Accessed 1 July 2022).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1095
(Fan et al., 2019; Quintero-Araujo et al., 2019; Rahmani et al., 2020). Under the horizontal coop-
eration relationship, EPSC members and LSSs suppliers would cooperate and compete with each
other simultaneously. For example, Tmall announced to cooperate with its rival Suning to build a
sharing platform.9
Motivated by the existence of different operation strategies in EPSCs, it is worth studying
whether and how to introduce LSSs. Therefore, by considering different operation strategies, we
construct three models of EPSC consisting of a manufacturer, an E-commerce platform, and
a LSSs supplier when LSSs are introduced. In this study, we investigate the following issues:
(1) Whether and how to introduce LSSs in the EPSC and decision-making of EPSC members
in different models? (2) What is the impact of the LSSs on the decisions of EPSC members and
the selection of operation strategy? (3) How to propose the optimal operation strategy by set-
ting the signing fee and controlling production cost to benefit the EPSC members, system, and
consumers?
Our contributions to the literature are three aspects. First, to the best of our knowledge, this
study is the first to theoretically investigate the operation and management of EPSCs by incorpo-
rating LSSs. Second, based on the impact of the LSSs on the live streaming channel, our work
goes further to consider the spillover effect that would influence the demands of the E-commerce
platform, which is rarely discussed in existing research. Finally, we propose a method of selecting
the operation strategy that is not only beneficial to the EPSC members, but also the EPSC system
and consumers.
The remainder of this paper is organized as follows. Section 2 presents the overview of the related
literature. Section 3 describes the problem and model assumptions. Section 4 gives the equilibrium
results of the three models in the EPSC. Section 5 analyzes and compares the equilibrium results
of the different three models. Section 6 summarizes the conclusions and gives some managerial
insights. All the proofs can be found in the Appendix.

2. Literature review

There are three main streams of literature closely related to this study: decision-making in EPSCs,
application of live streaming commerce, and operation strategy in supply chains.

2.1. Decision-making in EPSCs

The first stream of literature concerns the decision-making in EPSCs, which received considerable
attention (Shi et al., 2022; Zhang et al., 2022b). The majority of previous studies mainly focus on the
topics such as price strategy (Li et al., 2021b; Zhao et al., 2022), advertising investment (Du et al.,
2019), service level (Bian et al., 2021; Guo et al., 2021b), and spillover effects (Xia and Niu, 2019).
Specifically, Lu and Liu (2013) examined the impact of channel acceptance on the pricing strategy
and profits in an EPSC. Hao et al. (2017) analyzed the role of two unique features of advertising
in determining the optimal advertising revenue-sharing contract in an EPSC. Other studies related

9
Available at https://www.agoow.cn/xinwen/1217.html?ivk_sa=1024320u (Accessed 1 July 2022).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1096 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
to decision-making in EPSCs concentrated on the operation mode selection problem (Wang et al.,
2019). Jiang et al. (2011) found that platforms such as Amazon act as both a retailer and a reseller.
Thereafter, some scholars studied the platform selling mode (Qin et al, 2020), but many studies
consider only one selling mode and lack a comparison among different selling modes. The existing
literature pays more attention to the selling mode selection, and most of the literature is related to
different influence factors such as competition among members (Abhishek et al., 2016), consumer
perspective (Cao et al., 2020), and market size (Liu et al., 2020).
All of these studies focus on the decisions of pricing, service level, and advertising investment.
In contrast, this paper introduces the LSSs into an EPSC and investigates how the LSSs affect the
operation strategy selection for EPSC members.

2.2. Application of live streaming commerce

The second stream of related literature is about application of live streaming commerce (Park and
Lin, 2020; Lu et al., 2021; Zhang et al., 2022a). Live streaming is a type of user-generated content
(Lu and Chen, 2021), and it is also a new type of E-commerce shopping in which consumers can
interact with streamers in real time (Kang et al., 2021). The research mainly explored the factors
that influence consumer shopping behavior, including influencer effect (Hou et al., 2019), consumer
use intention (Chen and Lin, 2018) and consumer trust (Guo et al., 2021a). In particular, Zhuo
et al. (2019) evaluated the impact of social media influencers (SMIs) on the live streaming users.
Park and Lin (2020) explored the influences of various matches on the consumer attitudes with
the context of live streaming shopping with internet celebrities in China. The rise of live stream-
ing also boosted its fusion with platform (Kang et al., 2021; Li et al., 2021a). For example, Chen
and Lin (2018) studied the influence of platform-generated factors such as flow, social interac-
tion, and endorsement on customer behavior. Sun et al. (2019) found that live streaming shop-
ping platforms factors such as IT affordance play an important role in customer behavior. Li et al.
(2021c) developed a theoretical model to understand how LSSs affect user stickiness through users
attachment.
This area of research mainly empirically studies the application of live streaming commerce. Few
studies examine application issues of live streaming commerce from the perspective of operation
management. Accordingly, we examine whether and how to introduce LSSs in an EPSC, and how
the LSSs affect the decisions of pricing and service and operational performance of the EPSC
system. To the best of our knowledge, this study is the first to address these issues in EPSCs by
considering different operation strategies.

2.3. Operation strategy in supply chains

The third stream of literature relates to the operation strategy in supply chains, which aroused ex-
tensive attention (Qi et al., 2017; Dong et al., 2021). Many existing studies focused on channel
structure selection in supply chains (He et al., 2020). In particular, Tian et al. (2022) derived man-
agerial insights by considering horizontal differentiation across channels. Qiao and Su (2021) ad-
dressed the choice issues of a manufacturers licensing strategy and the independent remanufacturers

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1097
distribution channel. Xu and Choi (2021) analytically found that an increase of the cross-channel
effect and platform power can benefit the manufacturer and online platform. Xiao et al. (2020)
found that the traditional retail channel and the dual-channel could be the optimal choice for imple-
menting trade-ins. Some scholars concentrate on other operation strategies in supply chains, such
as pricing and service effort strategies (Bian et al., 2019), share-holding strategy (Ren et al., 2021a),
coordination strategy (Zhao et al., 2020), financing strategy (Yoo et al., 2021), product-design strat-
egy (Liu et al., 2019), and so on. For example, Ma (2021) designed an improvement contract by
considering the green tourism experience. Huang et al. (2020 c) investigated green credit, manufac-
turer subsidy, and sales subsidy modes in a supply chain consisting of a capital-constrained green
manufacturer. Zhou et al. (2021) investigated the price decisions and the conditions under which
an authorization agreement can be reached between the two parties.
These works mainly focused on the different operation strategies in supply chains. Our study
differs in two regards. First, we transform the subject from supply chains to EPSCs by considering
LSSs. Second, we introduce LSSs into an EPSC and investigate how the LSSs affect the selection
of operation strategy.
In summary, similar to Bell et al. (2018) and Du et al. (2019), this paper investigates the decisions
of price, service level, and advertising investment in an EPSC and proposes a method of selecting
operation strategy for EPSC members. We also explore whether and how to introduce LSSs, which
is similar to Guo et al. (2021a). Different from the above literature, we focus on the operation
strategy in an EPSC, and further discuss the impact of the LSSs on the decisions of EPSC members
and operation strategies. In addition, to enrich the related research, we not only comprehensively
investigate the EPSCs decision-making, but also explore how to put forward a consistent operation
strategy, which is beneficial for EPSC members, EPSC system, and consumers.

3. Problem description and model assumptions

3.1. Problem description

We consider an EPSC consisting of a manufacturer (M), an E-commerce platform (E), and a LSSs
supplier (L) when LSSs are introduced. In this EPSC, M sells products to consumers through plat-
form selling mode. By considering whether and how to introduce LSSs, three models are found:
benchmark model without LSSs (Model NL), operation model of LSSs introduced by M (Model
ML), and operation model of LSSs introduced by E (Model EL), which are illustrated in the fol-
lowing Fig. 1.
Similar to He et al. (2022), in the benchmark model NL, M sells products to consumers di-
rectly through E-commerce platform channel, and M pays a certain proportion of the revenue
and slotting fee to E. When the LSSs are introduced, M sells products to consumers through
E-commerce platform and live streaming channel by sharing a certain proportion of the revenue
with E and L (Xie et al., 2021), respectively. In the model ML, M introduces and adopts the LSSs
by paying a signing fee to L. While in the model EL, E introduces and adopts the LSSs by pay-
ing a signing fee to L (Fan et al., 2019; Guo et al., 2021b; Liu et al., 2021). For example, the
E-commerce platform Taobao launched Taobao live streaming to sell products (Hu and Chaudhry,
2020).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1098 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121

Fig. 1. Three models in the EPSC.

In Fig. 1, we can also find that, in the model ML, M shares a certain proportion of the revenue
from the live streaming channel with L, such as in Twitch.tv.10 In the model EL, L obtain a certain
proportion of M’s revenue and pay a certain proportion to E, such as Pinduoduo and Kuaishou.11
Moreover, the LSSs introduced by M or E has a service spillover effect (Chai et al., 2020),
which represents the impact of the LSSs on the E-commerce platform channel (Abhishek et al.,
2016).

3.2. Notation and assumptions

Based on the problem description, we employ the symbols and notations given in Table 1 through-
out this paper.
To make the analysis tractable, we introduce the following assumptions.
Consumer√demand. √ We assume that the positive effect of advertising/service investment on the
demand is ε ke /ε kl , ε > 0 (Du et al., 2019). Moreover, the market demand is a linear function
of the price and service level. Considering the advertising/service
√ factor and service spillover
√effect,
the demands of the two channels are De = α − βe pe + ε ke + ρs and Dl = α − βl pl + ε kl + s
(Ma, 2021; Qin et al., 2021), respectively, where α > 0, βl ≥ βe > 012 , and 0 ≤ ρ ≤ 1 (Xia and Liu,
2019; Chai et al., 2020).
Cost structure. Following Xue et al. (2020) and Zhang et al. (2021a), the service cost of live
streaming channel depends on the service level provided by L. Thus, the service investment cost

10
Available at https://www.sohu.com/a/127500647_483767 (Accessed 1 July 2022).
11
Available at https://zhuanlan.zhihu.com/p/262053824 (Accessed 1 July 2022).
12
http://www.ifastdata.com/data/upload/ueditor/20211009/616104925d62a.pdf (Accessed 13 April 2022).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1099
Table 1
Symbols and notations

Notations Definitions
Indexes
i EPSC member i, where i ∈ {E, L, M, T } and E, L, M, and T refer to the E-commerce
platform, LSSs supplier, manufacturer, and EPSC system, respectively
j Model j, where j ∈ {NL, ML, EL}
h Channel h, where h ∈ {e, l, t} and e, l, and t refer to the E-commerce platform channel, live
streaming channel, and two channels, respectively
Parameters
α Potential market size of the two channels
βe /βl Price elasticity coefficient of the E-commerce platform/live streaming channel
ε Consumers sensitivity coefficient toward advertising/service
ρ Degree of the service spillover effect
φ Revenue sharing proportion paid by M/L to E
λ Revenue sharing proportion paid by M to L
cm Unit production cost of products manufactured by M
ce /cl Unit operation cost of E-commerce platform/live streaming channel
Fe Slotting fee paid by M to E
Fl Signing fee paid by M/E to L
Decision variables
s Service level provided by L
ke /kl Investment of E’s advertising/L’s service
pe /pl Unit retail price of E-commerce platform/live streaming channel
Other notations
Dhj Demand of channel h in model j
πi j Profit of EPSC member i in model j

is assumed to be a convex function of the service level, that is, 12 s2 (Li et al., 2019b; Zhang et al.,
2021b). In addition, since E and L only display the products (Liu et al., 2021), while M produces
products and processes customer orders, we assume that the unit operation cost of E-commerce
platform or live streaming channel is zero, that is, ce = cl = 0 (Li et al., 2019b; Zhang et al., 2021a).
Power structure. Following Fan et al. (2019), Rahmani et al. (2020), and Quintero-Araujo et al.
(2019), both the cooperative and competitive relationships are assumed to exist simultaneously in
the EPSC. Meanwhile, we assume that the E and M have the highest and the lowest bargaining
powers, respectively (Xie et al., 2021; Shen et al., 2022). Thus, in the model NL, M and E play a
Stackelberg game, in which E is the leader and M is the follower. In the models ML and EL, M,
E, and L paly a sequential game, in which E is the leader, followed by L, and lastly by M (Zennyo,
2020; Zhang et al., 2021a).
If the price elasticity coefficient of the E-commerce platform channel βe is sufficiently large and
that of the live streaming channel βl is very small, that is, βe > 2α(1−φ)
cm
and βl < λ2 , and then the de-
mand of the E-commerce platform channel is negative, and thus, L will not provide service (Li et al.,
2019a, 2019b; Zhang et al., 2021a, 2021b). Therefore, to ensure that the optimal decision results are
meaningful and generate interesting managerial insights, we focus only on the case of βe ≤ 2α(1−φ) cm
and βl ≥ λ2 (which we elaborate on and demonstrate in the proofs of Theorems 1 and 2).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1100 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
4. Models and equilibrium results

In this section, we derive the equilibrium results for the models NL, ML, and EL. Following Zhang
et al. (2021b), we denote γ1 = 2βl − λ, γ2 = 2βl − λ + λφ, γ3 = 2βl − λ + 2ρλ, and γ4 = 2βl − λ +
λφ + 2ρ(1 − λ). According to the above assumptions of 0 < φ < 1 and βl ≥ λ2 , we can easily find
that γ2 , γ3 , γ4 > γ1 > 0, and γ4 > γ2 > 0.

4.1. Model NL

In the model NL, the LSSs are not introduced, M sells products directly to consumers through
E-commerce platform channel by paying slotting fee Fe to E and sharing a certain proportion φ
of the revenue with E. M and E play a dynamic Stackelberg game. E firstly decides the advertising
investment ke , and then M sets the retail price pe . The optimization models of profit maximization
for E and M are formulated as

max πENL = φ pNL

e De + Fe ,
NL
(1)
kNL
e

max πM
NL
= ((1 − φ)pNL
e − cm )De − Fe .
NL
(2)
pNL
e

By using the reverse induction, we have the optimal results for the model NL, which are described
in the following Theorem 1.
∗ α2
Theorem 1. In the model NL, the optimal advertising investment is kNL
e = ε2
, and the retail price of
NL∗ 2α(1−φ)+βe cm
the E-commerce platform channel is pe = 2βe (1−φ) .
And thus, the optimal demand of the E-commerce platform channel, profits of M, E, and
∗
NL∗ (2α(1−φ)−βe cm )2 NL∗ (4α 2 (1−φ)2 −βe2 c2m )φ
EPSC system are: DNL
e = 2α(1−φ)−β
2(1−φ)
e cm
, π M = 4βe (1−φ)
− Fe , π E = 4βe (1−φ)2
+ Fe ,
∗ ∗ ∗ ∗ 2α(1−φ)
and πTNL = πM NL
+ πENL . To ensure that DNL
e ≥ 0, we assume that βe ≤ cm
, and this assump-
tion holds throughout the paper (Ren et al., 2021b).

4.2. Model ML

In the model ML, the LSSs are introduced by M, and M sells products to consumers through
the E-commerce platform and live streaming channels by paying slotting fee Fe and signing fee Fl
to E and L, respectively. In addition, M shares a certain proportion φ or λ of revenue with E or
L, respectively. M, E, and L play a three-stage sequential game (Fan et al., 2019). E first decides
the advertising investment ke , and then L determines the service investment kl and service level s.
Finally, M sets the retail price pe and pl . The optimization models of profit maximization for E, L,
and M are formulated as

max πEML = φ pML ML

e De + Fe , (3)
kML
e

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1101
1
max πLML = λpML ML
l Dl + Fl − (sML )2 , (4)
kML
l ,sML 2

max πM
ML
= ((1 − φ)pML
e − cm )DML
e − Fe + ((1 − λ)pML
l − cm )DML
l − Fl . (5)
e ,pl
pML ML

By using the reverse induction, we have the optimal results for the model ML, which are described
in the following Theorem 2.
∗ α 2 γ32
Theorem 2. In the model ML, the optimal advertising and service investments are kML
e = ε 2 γ12
,
∗ α2 ∗
kML
l = ε2
, respectively; the service level is sML = 2λα
γ1
; and the retail prices of the two channels are
∗ α 2 γ32 +αγ3 ∗
pML
e = 2βe γ1
+ 2(1−φ)
cm
and pML
l = 2α
γ1
+ cm
2(1−λ)
.
∗
To ensure that sML ≥ 0, we assume that βl ≥ λ2 (Ren et al., 2021b), which holds throughout the
paper. Theorem 2 shows that, with the increase of the potential market demand of products, the
advertising and service investments, service level, and retail prices of the two channels all increase.
When the production cost increases, M will raise the retail prices of two channels.
And thus, the demands of E-commerce platform and live streaming channels, profits of M, E, L,
and EPSC system are

∗ 3αγ3 − α 2 γ32 βe cm ∗ 2αβl βl cm

DML
e = − , DML l = − ,
2γ1 2(1 − φ) γ1 2(1 − λ)
 2 2  
∗ (α γ3 + αγ3 )(1 − φ) cm 3αγ3 − α 2 γ32 βe cm
πM
ML
= − −
2βe γ1 2 2γ1 2(1 − φ)
  
2α(1 − λ) cm 2αβl βl cm
+ − − − Fe − Fl ,
γ1 2 γ1 2(1 − λ)
 2 2  
∗ α γ3 + αγ3 cm 3αγ3 − α 2 γ32 βe cm
πEML =φ + − + Fe ,
2βe γ1 2(1 − φ) 2γ1 2(1 − φ)
 2 
∗ 4α c2m 2α 2 λ2 ∗
ML∗ ∗ ∗
πLML = λβl − − + Fl , and πTML = πM + πEML + πLML .
γ12 4(1 − λ) 2 γ12

4.3. Model EL

In the model EL, the LSSs are introduced by E, and M sells products to consumers through the
live streaming and E-commerce platform channels. M and E pay slotting fee Fe and signing fee
Fl to E and L, respectively. M and L all share a certain proportion φ of the revenue with E. In
addition, M also shares a certain proportion λ of the revenue with L. M, E, and L play a three-
stage sequential game (Fan et al., 2019). E first decides the advertising investment ke , and then

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1102 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
L determines the service investment kl and service level s, and finally, M sets the retail prices pe and
pl . The optimization models of profit maximization for E, L and M are formulated as

max πEEL = φ(pEL

e De + λpl Dl ) + Fe − Fl ,
EL EL EL
(6)
kEL
e

1
max πLEL = λ(1 − φ)pEL EL
l Dl + Fl − (sEL )2 , (7)
kl ,s
EL EL 2

max πM
EL
= ((1 − φ)pEL
e − cm )De − Fe + ((1 − λ)pl
EL EL
− cm )DEL
l , (8)
e ,pl
pEL EL

By using the reverse induction, we have the optimal results for the model EL, which are described
in the following Theorem 3.
∗ α 2 γ42
Theorem 3. In the model EL, the optimal advertising and service investments are kEL
e = ε 2 γ22
and
∗ α2 ∗ 2λ(1−φ)α
kEL
l = ε2
, respectively; the service level is sEL = γ2
; and the retail prices of the two channels
∗ α 2 γ42 +αγ4 ∗
are pEL
e = 2βe γ2
+ cm
2(1−φ)
and pEL
l = 2α
γ2
+ cm
2(1−λ)
.

Theorem 3 shows that, with the increase of the potential market demand of the products, the
advertising, and service investments, service level and retail prices in the two channels all increase.
When the production cost increases, M will raise the retail prices of the two channels.
And thus, the optimal demands of the E-commerce platform and live streaming channels, profits
of M, E, L, and EPSC system are

∗ 3αγ4 − α 2 γ42 βe cm ∗ 2αβl βl cm

DEL
e = − , DELl = − ,
2γ2 2(1 − φ) γ2 2(1 − λ)
 2 2  
∗ (α γ4 + αγ4 )(1 − φ) cm 3αγ4 − α 2 γ42 βe cm
πM
EL
= − −
2βe γ2 2 2γ2 2(1 − φ)
  
2α(1 − λ) cm 2αβl βl cm
+ − − − Fe ,
γ2 2 γ2 2(1 − λ)
 2 2  
∗ α γ4 + αγ4 cm 3αγ4 − α 2 γ42 βe cm
πEEL =φ + −
2βe γ2 2(1 − φ) 2γ2 2(1 − φ)
  
2α cm 2αβl βl cm
+λφ + − + Fe − Fl ,
γ2 2(1 − λ) γ2 2(1 − λ)
∗ 4α 2 c2m 2α 2 λ2 (1 − φ)2 ∗
EL∗ ∗ ∗
πLEL = λ(1 − φ)βl ( − ) − + Fl , and πTEL = πM + πEEL + πLEL .
γ22 4(1 − λ)2 γ22

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1103
5. Characterization and comparison of equilibrium results

In this section, we explore the decisions of M, E, and L, and investigate the impact of market,
production, and revenue sharing parameters on the EPSC members decisions and profits. First, we
characterize the equilibrium results in Subsection 5.1. In Subsection 5.2, we compare the optimal
solutions of the three models. Finally, we derive the optimal operation strategy in Subsection 5.3.

5.1. Characterization of EPSC members’ optimal decisions

By analyzing the impacts of unit production cost cm , service spillover ρ, and revenue sharing pro-
portion λ on the retail prices, we have the following Proposition 1.
Proposition 1. The change trends of the retail prices of the two channels with parameters cm , ρ, and λ
are:
∗ ∗ ∗ ∗ ∗
(1) ∂ pNL
e /∂cm = ∂ pMLe /∂cm = ∂ pEL
e /∂cm > 0, ∂ pl
ML
/∂cm = ∂ pEL
l /∂cm > 0;
∗ ∗ ∗ ∗
(2) ∂ pe /∂ρ > ∂ pe /∂ρ > ∂ pl /∂ρ = ∂ pl /∂ρ = 0;
ML EL ML EL
∗ ∗
(3) ∂ pML
l /∂λ > ∂ pEL
l /∂λ > 0.

Proposition 1 (1) shows that a higher unit production cost will increase the retail prices of the
E-commerce platform and live streaming channels. That is to say, M will increase the retail prices
of two channels with the increase of production cost in the three models NL, ML, and EL. Ad-
ditionally, the change trends of the retail prices in three models are the same. Hence, production
cost has the same effect on pricing strategy in the three models. Proposition 1 (2) indicates that
when the LSSs are introduced, a higher service spillover effect coefficient will increase the retail
price of the E-commerce platform channel, which will not affect the retail price of the live stream-
ing channel. And the increase range of the retail price in the model ML is greater than that in the
model EL. Because the service spillover effect can expand market share. Therefore, M should in-
troduce and adopt the LSSs to heighten the service spillover effect to expand market share, and
then take advantage of this effect to obtain more benefits. Proposition 1 (3) shows that, when
the LSSs are introduced, with the increase of the revenue sharing proportion paid by M to L,
M will increase the retail prices of the live streaming channel, and the increase range of the retail
price in the model ML is greater than that in the model EL. Because when M shares its revenue
with L, M’s revenue decreases, and then M will heighten the retail prices of the live streaming
channel.
By analyzing the effects of potential market demand α and the service spillover effect coefficient
ρ on the advertising investment, we have the following Proposition 2.
Proposition 2. The change trends of the advertising investment with parameters α and ρ are:
∗ ∗ ∗
e ∗ /∂α > ∂ke ∗ /∂α > ∂ke
(1) ∂kML EL NL
/∂α > 0;
(2) ∂ke /∂ρ > ∂ke /∂ρ > 0.
ML EL

From the perspective of advertising investment, Proposition 2 indicates that higher potential mar-
ket demand and service spillover effect will increase the advertising investment. This implies that

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1104 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
EPSC members should invest advertising to expand market share and then heighten the service
spillover effect. Specifically, the effect of potential market demand on the advertising investment in
the model ML is the highest, while this effect in the model NL is the lowest. The service spillover
effect on the advertising investment in the model ML is higher than that in the model EL. Conse-
quently, it can be suggested that manufacturers should introduce and adopt the LSSs to increase
the advertising investment and thus expand more potential consumers.
By analyzing the impacts of potential market demand α, service investment sensitivity coeffi-
cient ε, and revenue sharing proportion λ on the service investment and service level, we have the
following Proposition 3.
Proposition 3. The change trends of the service investment and service level with α, ε, and λ parameters
are:
∗ ∗ ∗ ∗
(1) ∂kML
l /∂α = ∂kEL
l /∂α > 0, ∂kML
l /∂ε = ∂kEL
l ∗ /∂ε < 0;
ML∗ EL∗ ML∗
(2) ∂s /∂α > ∂s /∂α > 0, ∂s /∂λ > ∂sEL /∂λ > 0.

Proposition 3 (1) indicates that a higher potential market demand will increase the service invest-
ment, but a higher service investment sensitivity will decrease the service investment. In addition,
the effects of potential market demand and service investment sensitivity on the service investment
in the models ML and EL are the same. This implies that EPSC members should introduce and
adopt the LSSs to attract more potential consumers, and then motivate L to increase the service
investment. Proposition 3 (2) shows that a higher potential market demand and revenue sharing
proportion will increase the service level, and the effects of potential market demand or revenue
sharing proportion on the service level in the model ML are larger than that in the model EL.
From the perspective of increasing L’s service level, it can be suggested that EPSC members should
invest advertising to expand market share and set an appropriate revenue sharing proportion.

5.2. Comparison of EPSC members’ optimal decisions

In this part, by comparing the equilibrium results in the three models NL, ML, and EL, we can
derive the following conclusions.
First, by comparing the optimal retail prices in the three models, we have the following Proposi-
tion 4.
Proposition 4. The optimal retail prices of the two channels in the different three models satisfy
∗ ∗ ∗ ∗
(1) if α > 1, then pML
e > pNLe , otherwise pMLe ≤ pNL
e ;
EL∗ NL∗ EL∗ NL∗
if α > 2 , then p e > p e ∗, otherwise p e ∗ ≤ p e ∗;
EL∗
if α > 3 , then pe > pML
e , otherwise pEL
e ≤ pML
e ;
∗ ∗
(2) pML
l > pEL
l .

Proposition 5 (1) indicates that when the potential market demand is high enough, the retail
prices of the E-commerce platform channel in the models ML and EL are all higher than that in
the model NL, while the retail price of the E-commerce platform channel in the model EL is higher
than that in the model ML. However, only the thresholds of potential market demand are different

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1105

Fig. 2. Impact of potential market demand on the retail price.

in each model (see Fig. 2). In other words, under some conditions, M will set a higher retail price
when the LSSs are introduced, and the retail price is the highest in the model EL. This implies
that M will set an appropriate retail price of the E-commerce platform channel according to the
parameters of the market, revenue sharing proportion, and service spillover effect.
Different from the retail price of the E-commerce platform channel, when the LSSs are intro-
duced, Proposition 4 (2) shows that the retail price of the live streaming channel in the model ML
is higher than that in the model EL. That is to say, M sets higher retail price of the live streaming
channel in the model ML than that in the model EL. The reason is that, M can control the live
streaming channel in the model ML, and then set a higher retail price of the live streaming channel.
To describe the impacts of potential market demand on the retail prices of E-commerce platform
channel in the three models, the parameters are setting as βe = 0.1, βl = 0.12, ρ = 0.1, ε = 0.1,
φ = 0.2, λ = 0.1, cm = 5, Fe = 500, and Fl = 1000. And then, the thresholds of 1 , 2 , and 3
are calculated as 1 = 4.6875, 2 = 4.6488, and 3 = 3.9894. To clearly describe the relationship
between the optimal retail prices in the E-commerce platform channel, α varied within the range of
[3.8,4.8]. Figure 2 illustrates the optimal retail prices in three models.
Figure 2(a) indicates that a higher potential market demand will increase the retail price of the E-
commerce platform channel. Besides, when the potential market demand is enough high, M will set
the highest retail price of the E-commerce platform channel in the model EL. On the contrary, when
the potential market demand is low, M will set the highest retail price of the E-commerce platform
channel when the LSSs are not introduced. This implies that EPSC members should introduce and
adopt the LSSs to explore more potential consumers. Manufacturers determine the retail prices
according to the parameters of the market, cost, and revenue sharing proportion. Figure 2(b) shows
that, a higher potential market demand will increase the retail price of the live streaming channel.
M always sets a higher retail price of the live streaming channel in the model ML than that in the
model EL, which verifies the conclusions in the Proposition 4.
We set α = 10, βl ∈ [0.1, 0.15], and ρ ∈ [0.1, 0.3], and illustrate the effect of the price elasticity
coefficient of live streaming channel βl and service spillover effect ρ on the retail price in the three
models, which are described in the following Fig. 3.

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1106 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121

Fig. 3. Impact of retail price elasticity and service spillover effect on the retail price.

From Fig. 3(a), we can find that a higher service spillover effect will increase the retail price of
the E-commerce platform channel. When the effects of the price elasticity and service spillover on
the retail price of E-commerce platform channel are very low, M will set higher retail price of the
E-commerce platform channel when the LSSs are introduced. When the service spillover effect is
high, the retail price of the E-commerce platform channel in the model ML is higher than that in
the model EL. This implies that the LSSs and service spillover effect can raise the retail price of
the E-commerce platform channel. Therefore, manufacturers would more likely to introduce and
adopt the LSSs from LSSs supplier. Figure 3(b) indicates that M will decrease the retail price of
the live streaming channel with the increase in the price elasticity coefficient of the live streaming
channel. The service spillover effect will not affect the M’s price decisions. These findings verify the
conclusions in the Propositions 1 and 4.
Next, we analyze the optimal decisions of E and L in the three models. By comparing the op-
timal advertising investments, service investments, and service levels in the three models, we have
Proposition 5.
Proposition 5. The optimal advertising investment, service investment, and service level in the three
models satisfy
∗ ∗ ∗
e ∗ > ke ∗ > ke
(1) kML EL NL
;
ML
(2) kl = kl ;
EL
∗ ∗
(3) sML > sEL .

Proposition 5 (1) indicates that when the LSSs are introduced, M will increase its advertising in-
vestment. Additionally, E will provide the highest advertising investment in the model ML. There-
fore, M and E should agree to introduce the LSSs to increase the adverting investment and attract
more potential consumers. Proposition 5 (2) shows that when the LSSs are introduced, the LSSs
supplier will provide the same service investments in the models ML and EL. Proposition 5 (3)
shows that the L will provide a higher service level in the model ML than that in the model EL.

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1107

Fig. 4. Impact of potential market demand on demand.

Consequently, EPSC members should introduce the LSSs, and the operation strategy of the LSSs
introduced by M will increase the advertising investment and service level, which can expand mar-
ket share.
By denoting Dti = Die + Dil , i ∈ {NL, ML, EL}, and comparing the equilibrium demands of the
E-commerce platform channel, live streaming channel, and two channels in the three models, we
have the following Proposition 6.
Proposition 6. The demands of the two channels in the three models satisfy:
∗ ∗ ∗ ∗
e ∗ > De ∗ , otherwise De ∗ ≤ De ∗ ;
(1) if α < 4 , then DML NL ML NL
∗ ∗
if α < 5 , then De > De , otherwise De ≤ De ; DML
EL NL EL NL
e > DEL
e ;
∗ ∗
(2) DMLl > DEL
l ;
∗ ∗ ∗ ∗
(3) if 6 < α < 7 , then DtML > DtNL , otherwise DtML ≤ DtNL ;
∗ ∗ ∗ ∗ ∗ ∗
if 8 < α < 9 , then DtEL > DtNL , otherwise DtEL ≤ DtNL ; DtML > DtEL .

Proposition 6 (1) indicates that when the potential market demand is low, comparing with the
case without LSSs, the LSSs will increase the demand of the E-commerce platform channel in
the models ML and EL. However, only the thresholds of potential market demand in the three
models are different (see Fig. 4). Specifically, the demand of the E-commerce platform channel in
the model ML is larger than that in the model EL. Proposition 6 (2) shows that the demand of the
live streaming channel in the model ML is larger than that in the model EL. Hence, the LSSs should
be introduced by M, which can increase the market demand. Proposition 6 (3) shows that when the
potential market demand satisfies some conditions, the total demands in the models ML and EL
are all higher than that in the model NL. When the potential market demand is either very low or
high, the total demand in the model NL is higher than that in the model ML or EL. In other words,
the LSSs do not always increase the total demand of the EPSC. When the LSSs are introduced, the
total demand in the model ML is higher than that in the model EL. Therefore, manufacturers

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1108 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121

Fig. 5. Impact of price elasticity and service spillover effect on demand.

should introduce and adopt the LSSs to expand the market demands of each channel and the total
demand of EPSC.
Under the above parameter settings, the thresholds of 4 − 9 are calculated as 4 = 7.8125,
5 = 6.7149, 6 = 0.1380, 7 = 26.4245, 8 = 0.1561, and 9 = 22.0546. To clearly describe the
relationships of the optimal demands of the E-commerce platform channel and total demand in
the three models, α varied within the range of [6,8] and [0,30], respectively. Figure 4 describes the
optimal demands in the three models.
Figure 4 indicates that with the increase of the potential market demand, the demand of the E-
commerce platform channel and the total demand of EPSC in the model NL all increase, but the
demand of the E-commerce platform channel and the total demand of EPSC in the models ML
and EL first increase and then decrease. In addition, when the potential market demand is low, the
demand of the E-commerce platform channel and the total demand of EPSC in the model ML are
the highest. When the potential market demand is high, the demand of the E-commerce platform
channel and the total demand of EPSC in the model NL are the highest. And thus, EPSC members
should introduce and adopt the LSSs, invest advertising, etc., to explore the potential demand of
the market to derive greater profits. Moreover, when the market demand is close to saturation, there
is no need to introduce the LSSs. When the LSSs are introduced, comparing with the case in the
model EL, it is more conducive to expanding the market demands of the E-commerce platform and
live streaming channels and the total demand of EPSC in the model ML. That is to say, to increase
the market demand, the LSSs should be introduced by manufacturers. These verify the conclusions
in the Proposition 6.
To ensure that the demand is positive, we set α = 10, βl ∈ [0.1, 0.15], and ρ ∈ [0.1, 0.3]. By an-
alyzing the effects of price elasticity coefficient βl and service spillover effect ρ on the demand of
the E-commerce platform channel and the total demand of EPSC in the three models, we have the
following Fig. 5.
Figure 5 indicates that when the price elasticity is low, the demand of the E-commerce platform
channel in the model ML is the highest. When the price elasticity is sufficiently high, the demand

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1109
of E-commerce platform channel in the model NL is the highest. In addition, the total demand of
EPSC in the model ML is the highest. That is to say, the LSSs do not always increase demand of the
E-commerce platform channel, but it exploits the demand of the live streaming channel. Therefore,
under some conditions, the LSSs greatly expand the total market demand, and the operation strat-
egy of the LSSs introduced by M is more effective than of the LSSs introduced by E on expanding
the total demand of EPSC. Therefore, in the EPSC, the LSSs should be introduce by M, which can
expand more demands.

5.3. Optimal operation strategy for the EPSC

Comparing the optimal profits of the EPSC members, we derive the optimal operation strategy for
the EPSC in the following Propositions.
Proposition 7. The optimal profits of M in the three models satisfy:
∗ ∗ ∗ ∗
(1) if Fl < F1 , then πM
ML
> πM
NL
, otherwise πM
ML
≤ πM
NL
;
∗ ∗
EL∗ NL∗
(2) if 0 < cm < C1 or cm > C2 , then πM > πM , otherwise πM
EL NL
≤ πM ;
∗ ∗ ∗ ∗
(3) if Fl > F2 , then πM
EL
> πM
ML
, otherwise πM
EL
≤ πM
ML
.

Proposition 7 shows that when the signing fee is below a certain value, M will obtain more profit
in the model ML than that in the model NL. When the production cost is in a certain range, M’s
profit in the model EL is higher than that in the model NL. That is to say, when the signing and
production cost satisfy some conditions, M will prefer to introduce the LSSs. In addition, when the
signing fee is small, the operation strategy of the LSSs introduced by M is the best choice for M.
Otherwise, M will prefer E to introduce the LSSs. Therefore, manufacturers should control their
production costs, and then introduce and adopt the LSSs to expand market share when the signing
fee is small.
Proposition 7 gives the condition under which M can benefit from introducing the LSSs. Under
the above parameter settings, the thresholds of F1 , F2 , C1 , and C2 are calculated as F1 = 2262.9003,
F2 = 626.3608, C1 = 161.7161, and C2 = 318.2839. We employ Fig. 6 to illustrate the operation
strategy for M and the corresponding insights.
Figure 6(a) indicates that a higher signing fee will decrease M’s profit in the model ML. Since
cm = 5 < C1 = 161.7161, M’s profit in the model EL is always higher than that in the model NL. In
addition, when the signing fee is very small, M gains the highest profit in the model ML. While when
the signing fee is high, M’s profit in the model EL is the highest. That is to say, when the production
cost is in a certain range, the LSSs should be introduced and implemented in the EPSC. And when
the signing fee is small, the LSSs should be introduced by M. In other words, the operation strategy
of the LSSs introduced by M is the best choice for M. Otherwise, the operation strategy of the LSSs
introduced by E is beneficial for M. And thus, manufactures should control their production costs
and then introduce and adopt the LSSs. These verify the conclusions in Proposition 7.
From Fig. 6(b), we can find that, when the signing fee is very low, M prefers to introduce and
adopt the LSSs. When the signing fee is sufficiently high (i.e., Fl > F2 ) and the production cost is
low (i.e., cm < C1 ), the operation strategy of the LSSs introduced by E is beneficial for M. When
the signing fee and the product cost are high (i.e., Fl > F1 and cm > C1 ), M will not introduce

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1110 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121

Fig. 6. Optimal operation strategy for M.

the LSSs. Consequently, when the signing fee and production cost are sufficiently low or high,
M has different operation strategy preferences. Under different conditions, all the three operation
strategies may be selected by manufacturers. Therefore, manufacturers can control their production
costs, and the EPSC members can set an appropriate signing fee Fl to choose the optimal operation
strategy.
Proposition 8. The optimal profits of E in the three models satisfy:
∗ ∗ ∗ ∗
(1) if cm > C3 , then πEML > πENL , otherwise πEML ≤ πENL ;
∗ ∗ ∗ ∗
(2) if Fl < F3 , then πEEL > πENL , otherwise πEEL ≤ πENL ;
∗ ∗ ∗ ∗
(3) if Fl < F4 , then πEEL > πEML , otherwise πEEL ≤ πEML .

Proposition 8 indicates that when the production cost is above a certain value, E can obtain more
profit in the model ML than that in the model NL. In addition, when the signing fee is small, E’s
profit in the model EL is higher than that in the model NL or ML. That is to say, when the signing
fee is in a certain range, E would like to introduce and adopt the LSSs to obtain more profits. When
the production cost and signing fee satisfy some conditions, the operation strategy of the LSSs
introduced by E is beneficial for E.
Proposition 8 gives the condition under which the retailer can benefit from introducing the LSSs.
Under the above parameter settings, the thresholds of F3 , F4 , and C3 are calculated as F3 = 39.3117,
F4 = 5.8831, and C3 = −267.3684. We employ Fig. 7 to illustrate the operation strategy for E and
the corresponding insights.
In Fig. 7(a), we can find that, a higher signing fee will decrease E’s profit in the model EL.
Since cm = 5 > C3 = −267.3684, E can always obtain more profit in the model ML than that in
the model NL. In addition, when the signing fee is very small, the operation strategy of the LSSs
introduced by E is always beneficial for E. Otherwise, the operation strategy of the LSSs introduced
by M is beneficial for E. That is to say, the LSSs are beneficial for E in most cases. The reason is

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1111

Fig. 7. Optimal operation strategy for E.

that the LSSs can exploit more potential consumers. Therefore, E should introduce and adopt the
LSSs when the signing fee is sufficiently small. And thus, the EPSC members should negotiate to
determine the signing fee to choose the appropriate operation strategy.
Figure 7(b) indicates that when the signing fee is relatively low (i.e., Fl < F4 ), E prefers to intro-
duce and adopt the LSSs. Otherwise, the operation strategy of the LSSs introduced by M is the best
choice for E. In most cases, the operation strategy of the LSSs introduced by M is beneficial for
E. Therefore, the EPSC members should set an appropriate signing fee Fl to ensure that all EPSC
members choose the same operation strategy.
∗ ∗
Proposition 9. The optimal profits of L in the three models satisfy: if cm > C4 , then πLEL > πLML ,
∗ ∗
otherwise πLEL ≤ πLML .

Proposition 9 indicates that, when the production cost is high, L’s profit in the model EL is higher
than that in the model ML. This result implies that the LSSs supplier prefers E to introduce the
LSSs when the production cost is high. Otherwise, the LSSs supplier will prefer M to introduce
the LSSs. Therefore, manufacturers should control their production costs to introduce and adopt
the LSSs.
Proposition 9 gives the condition for L to select the operation strategy. Under the above param-
eter settings, the threshold of C4 is calculated as C4 = 240.5351. We employ Fig. 8 to illustrate the
operation strategy for L and the corresponding insights.
Figure 8(a) shows that a higher production cost will decrease the L’s profit. When the produc-
tion cost is low, L prefers M to introduce the LSSs. Otherwise, the operation strategy of the LSSs
introduced by E is beneficial for L. These results confirm that manufacturers should control their
production costs, and then the LSSs suppliers will prefer M to introduce the LSSs. Figure 8(b) in-
dicates that the signing fee will not affect L’s operation strategy choice. When the production cost is
low (i.e., cm < C4 ), the LSSs supplier prefers M to introduce the LSSs. Otherwise, the LSSs supplier

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1112 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121

Fig. 8. Optimal operation strategy for L.

Fig. 9. Profit of the EPSC system.

will prefer E to introduce the LSSs. Consequently, manufacturers should control their production
costs to introduce the LSSs.
Under the above parameter settings, to ensure that the profit of the EPSC system is positive, cm
varied within the range [0,100]. In addition, we set cm to be 5, βl and ρ varied within the ranges of
[0.1,0.15] and [0.1,0.3], respectively. The optimal profits of the EPSC in three models are described
in the following Fig. 9.
Figure 9 indicates that the EPSC’s profit in the model ML is the highest, while it is the lowest in
the model NL. This implies that introducing the LSSs are always beneficial for the EPSC system
because the LSSs can exploit more potential consumers, which can bring more benefits for EPSC

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1113

Fig. 10. Operation strategy for the EPSC system.

members. To be more specific, the operation strategy of the LSSs introduced by M is more effective
for EPSC system than that of the LSSs introduced by E. From the perspective of the EPSC system,
the operation strategy of the LSSs introduced by M is the best choice. Therefore, the EPSC members
should agree to let manufacturers introduce the LSSs by setting appropriate parameters.
To determine the conditions under which the EPSC members choose a consistent operation strat-
egy, we employ the above parameter settings and use Fig. 10 to illustrate the corresponding insights.
In Fig. 10, we can find that, when the signing fee and production cost parameters satisfy some
conditions (i.e., if cm < C1 , F4 < Fl < F2 and cm > C1 , F4 < Fl < F1 ), the operation strategy of
the LSSs introduced by M is the best choice for the EPSC members and system. In addition, with
the increase of the production cost, the value space of the consistent operation strategy becomes
smaller. Consequently, we can conclude that when the EPSC members set an appropriate signing
fee and manufacturers control their production costs, all EPSC members will choose the same
operational strategy of the LSSs introduced by M. In practice, many manufacturers would not only
sell products through E-commerce platform channel, but also choose to introduce LSSs to attract
more consumers. Manufacturers such as Gree and Shell prefer to directly introduce LSSs from the
live streaming platforms Kuaishou and TikTok to further expand the consumer demand.13,14

6. Conclusions

This study explores whether and how to introduce LSSs and puts forward an operation strategy in
an EPSC. We construct three models to analyze the decisions of EPSC members and investigate
the impacts of the LSSs on decisions and operation strategy. The results show that: (1) When the
production cost is very high, the manufacturer will not introduce the LSSs, while introducing the
LSSs is always beneficial for the E-commerce platform. However, the LSSs may not always increase

13
Available at https://www.sohu.com/a/407968919_120013420 (Accessed 1 July 2022).
14
Available at https://baijiahao.baidu.com/s?id=1669168968083401575&wfr=spider&for=pc (Accessed 1 July 2022).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1114 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
the demand of the E-commerce platform channel and the total demand of EPSC system. (2) When
the LSSs are introduced, the E-commerce platform and LSSs supplier should offer high levels of
adverting and service investment to expand more market demands. (3) EPSC members prefer dif-
ferent operation strategies under different conditions. When the signing fee and production costs
are relatively low, all three EPSC members will consistently choose the operation strategy of the
LSSs introduced by manufacturer, which is also beneficial for EPSC system and consumers.
Based on these findings, some management implications are given. (1) The EPSC members would
like to introduce the LSSs to the EPSC. However, the choice of consistency operation strategy
is conditional. Only when the signing fee and production costs are relatively low, manufacturers,
E-commerce platforms, and LSSs suppliers prefer the operation strategy of the LSSs introduced
by manufacturers. This is successfully practiced by the Chinese company Gree. As a traditional
manufacturer in China, Gree has gained ¥47.6 billion through live streaming channel in 2020.15
And thus, manufacturers and E-commerce platforms should negotiate an appropriate signing fee
to achieve a win–win situation. (2) To attract more potential consumers, the E-commerce platforms
and LSSs suppliers should offer high levels of adverting and service investment. For example, to
inform consumers of activities about products, E-commerce platforms should post some advertising
notices on their online stores and LSSs suppliers should promote live streaming events online. It is
also successfully practiced by some E-commerce platforms and LSSs suppliers such as TikTok and
Luo Yonghao16 . Therefore, EPSC members should set appropriate parameters to implement the
operation strategy of the LSSs introduced by manufacturers, which is the best choice for the EPSC
members, EPSC system, and consumers.
Our study could be extended in several ways in future research. We consider the impact of live
streaming on manufacturers and E-commerce platforms. In reality, some enterprises such as Gree
not only act as manufacturers, but also operate their own live streaming channel. Different channel
structure modes will be investigated in the future research. In addition, we regard the LSSs supplier
as only a participant in the ESPC. However, LSSs suppliers include live streaming platforms and
live streamers. In the future research, it is worth to explore operation strategy in EPSCs consisting
of different types of LSSs suppliers.

Acknowledgment

This research was supported by the National Natural Science Foundation of China (Grant No.
71771055 and No. 71801003).

References

Abhishek, V., Jerath, K., Zhang, Z.J., 2016. Agency selling or reselling? Channel structures in electronic retailing. Man-
agement Science 62, 8, 2259–2280.

15
Available at https://baijiahao.baidu.com/s?id=1704769940414445343&wfr=spider&for=pc (Accessed 1 July 2022).
16
Available at https://www.jiemian.com/article/4216218.html (Accessed 1 July 2022).

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1115
Agrawal, A.K., Yadav, S., 2020. Price and profit structuring for single manufacturer multi-buyer integrated inventory
supply chain under price-sensitive demand condition. Computers & Industrial Engineering 139, 106208.
Bell, D.R., Gallino, S., Moreno, A., 2018. Offline showrooms in omni-channel retail: demand and operational benefits.
Management Science 64, 4, 1629–1651.
Bian, Y., Cui, Y., Yan, S., Han, X., 2021. Optimal strategy of a customer-to-customer sharing platform: whether to launch
its own sharing service? Transportation Research Part E: Logistics and Transportation Review 149, 102288.
Bian, Y., Xie, J., Archibald, T.W., Sun, Y., 2019. Optimal extended warranty strategy: offering trade-in service or not?
European Journal of Operational Research 278, 1, 240–254.
Cao, K., Xu, Y., Cao, J., Xu, B., Wang, J., 2020. Whether a retailer should enter an e-commerce platform taking into
account consumer returns. International Transactions in Operational Research 27, 6, 2878–2898.
Chai, L., Duan, Y., Huo, J., 2021. Pricing strategies for O2O business model considering service spillover and power
structures. International Transactions in Operational Research 28, 4, 1978–2001.
Chen, C.C., Lin, Y.C., 2018. What drives live-stream usage intention? the perspectives of flow, entertainment, social
interaction, and endorsement. Telematics and Informatics 35, 1, 293–303.
Dong, C., Chen, C., Shi, X., Ng, C.T., 2021. Operations strategy for supply chain finance with asset-backed securitization:
centralization and blockchain adoption. International Journal of Production Economics 241, 108261.
Du, S., Wang, L., Hu, L., Zhu, Y., 2019. Platform-led green advertising: promote the best or promote by performance.
Transportation Research Part E: Logistics and Transportation Review 128, 115–131.
Fan, X., Yin, Z., Liu, Y., 2020. The value of horizontal cooperation in online retail channels. Electronic Commerce
Research and Applications 39, 100897.
Guan, L., Chen, H., Ma, H., Zhang, L., 2022. Optimal group-buying price strategy considering the information-sharing
of the seller and buyers in social e-commerce. International Transactions in Operational Research 29, 3, 1769–1790.
Guo, L., Hu, X., Lu, J., Ma, L., 2021a. Effects of customer trust on engagement in live streaming commerce: mediating
role of swift guanxi. Internet Research 31, 5, 1718–1744.
Guo, X., Zheng, S., Yu, Y., Zhang, F., 2021b. Optimal bundling strategy for a retail platform under agency selling.
Production and Operations Management 30, 7, 2273–2284.
Hao, L., Guo, H., Easley, R.F., 2017. A mobile platform’s in-app advertising contract under agency pricing for app sales.
Production and Operations Management 26, 2, 189–202.
He, P., He, Y., Tang, X., Ma, S., Xu, H., 2022. Channel encroachment and logistics integration strategies in an e-commerce
platform service supply chain. International Journal of Production Economics 244, 108368.
He, P., He, Y., Xu, H., 2020. Buy-online-and-deliver-from-store strategy for a dual-channel supply chain considering
retailers location advantage. Transportation Research Part E: Logistics and Transportation Review 144, 102127.
Hou, F., Guan, Z., Li, B., Chong, A., 2019. Factors influencing people’s continuous watching intention and consumption
intention in live streaming: evidence from china. Internet Research 30, 1, 141–163.
Hu, M., Chaudhry, S.S., 2020. Enhancing consumer engagement in e-commerce live streaming via relational bonds.
Internet Research 30, 3, 1019–1041.
Huang, S., Fan, Z.P., Wang, N., 2020. Green subsidy modes and pricing strategy in a capital-constrained supply chain.
Transportation Research Part E: Logistics and Transportation Review 136, 101885.
Jiang, B., Jerath, K., Srinivasan, K., 2011. Firm strategies in the “mid tail” of platform-based retailing. Marketing Science
30, 5, 757–775.
Kang, K., Lu, J., Guo, L., Li, W., 2021. The dynamic effect of interactivity on customer engagement behavior through
tie strength: evidence from live streaming commerce platforms. International Journal of Information Management 56,
102251.
Li, G., Li, L., Sethi, S.P., Guan, X., 2019a. Return strategy and pricing in a dual-channel supply chain. International
Journal of Production Economics 215, 153–164.
Li, G., Li, L., Sun, J., 2019b. Pricing and service effort strategy in a dual-channel supply chain with showrooming effect.
Transportation Research Part E: Logistics and Transportation Review 126, 32–48.
Li, R., Lu, Y., Ma, J., Wang, W., 2021a. Examining gifting behavior on live streaming platforms: an identity-based
motivation model. Information & Management 58, 6, 103406.
Li, Y., Li, B., Zheng, W., Chen, X., 2021b. Reveal or hide? impact of demonstration on pricing decisions considering
showrooming behavior. Omega 102, 102329.

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1116 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
Li, Y., Li, X., Cai, J., 2021c. How attachment affects user stickiness on live streaming platforms: a socio-technical ap-
proach perspective. Journal of Retailing and Consumer Services 60, 102478.
Liu, W., Liang, Y., Tang, O., Shi, V., Liu, X., 2021. Cooperate or not? Strategic analysis of platform interactions consid-
ering market power and precision marketing. Transportation Research Part E: Logistics and Transportation Review
154, 102479.
Liu, W., Yan, X., Li, X., Wei, W., 2020. The impacts of market size and data-driven marketing on the sales mode selection
in an internet platform based supply chain. Transportation Research Part E: Logistics and Transportation Review 136,
101914.
Liu, Z., Li, K.W., Li, B.Y., Huang, J., Tang, J., 2019. Impact of product-design strategies on the operations of a closed-
loop supply chain. Transportation Research Part E: Logistics and Transportation Review 124, 75–91.
Lu, B., Chen, Z., 2021. Live streaming commerce and consumers purchase intention: an uncertainty reduction perspec-
tive. Information & Management 58, 7, 103509.
Lu, Q., Liu, N., 2013. Pricing games of mixed conventional and e-commerce distribution channels. Computers & Industrial
Engineering 64, 1, 122–132.
Ma, P., 2021. Optimal generic and brand advertising efforts in a decentralized supply chain considering customer surplus.
Journal of Retailing and Consumer Services 60, 102502.
Ma, S., He, Y., Gu, R., 2021. Joint service, pricing and advertising strategies with tourists green tourism experience in a
tourism supply chain. Journal of Retailing and Consumer Services 61, 102563.
Park, H.J., Lin, L.M., 2020. The effects of match-ups on the consumer attitudes toward internet celebrities and their
live streaming contents in the context of product endorsement. Journal of Retailing and Consumer Services 52,
101934.
Qi, Y., Huo, B., Wang, Z., Yeung, H.Y.J., 2017. The impact of operations and supply chain strategies on integration and
performance. International Journal of Production Economics 185, 162–174.
Qiao, H., Su, Q., 2021. Distribution channel and licensing strategy choice considering consumer online reviews in a
closed-loop supply chain. Transportation Research Part E: Logistics and Transportation Review 151, 102338.
Qin, J., Fu, H., Wang, Z., Xia, L., 2021a. Financing and carbon emission reduction strategies of capital-constrained
manufacturers in e-commerce supply chains. International Journal of Production Economics 241, 108271.
Qin, X., Liu, Z., Tian, L., 2021b. The optimal combination between selling mode and logistics service strategy in an
e-commerce market. European Journal of Operational Research 289, 2, 639–651.
Quintero-Araujo, C.L., Gruler, A., Juan, A.A., Faulin, J., 2019. Using horizontal cooperation concepts in integrated
routing and facility-location decisions. International Transactions in Operational Research 26, 2, 551–576.
Rahmani, S., Haeri, A., Hosseini-Motlagh, S.M., 2020. Proposing channel coordination and horizontal cooperation in
two competitive three-echelon reverse supply chains. International Transactions in Operational Research 27, 3, 1447–
1477.
Ren, D., Guo, R., Lan, Y., Shang, C., 2021a. Shareholding strategies for selling green products on online platforms in a
two-echelon supply chain. Transportation Research Part E: Logistics and Transportation Review 149, 102261.
Ren, M., Liu, J., Feng, S., Yang, A., 2021b. Pricing and return strategy of online retailers based on return insurance.
Journal of Retailing and Consumer Services 59, 102350.
Shen, J., Qian, J., Shao, X., 2022. Service channel choice for supply chain under online marketplace. International Trans-
actions in Operational Research 29, 4, 2401–2422.
Shi, G., Jiang, L., Wang, Y., 2022. Interaction between the introduction strategy of the third-party online channel and
the choice of online sales format. International Transactions in Operational Research 29, 4, 2448–2493.
Singh, S., Singh, N., Kalinić, Z., Liébana-Cabanillas, F.J., 2021. Assessing determinants influencing continued use of live
streaming services: an extended perceived value theory of streaming addiction. Expert Systems with Applications
168, 114241.
Sun, Y., Shao, X., Li, X., Guo, Y., Nie, K., 2019. How live streaming influences purchase intentions in social commerce:
an it affordance perspective. Electronic Commerce Research and Applications 37, 100886.
Tian, C., Xiao, T., Shang, J., 2021. Channel differentiation strategy in a dual-channel supply chain considering free riding
behavior. European Journal of Operational Research 30, 473–485.
Tsunoda, Y., Zennyo, Y., 2021. Platform information transparency and effects on third-party suppliers and offline retail-
ers. Production and Operations Management 30, 11, 4219–4235.

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1117
Wang, T.Y., Wang, X.J., He, P., Chen, Z.S., 2022. Retailer’s selling mode choice under different competition forms. Inter-
national Transactions in Operational Research 29, 3712–3736.
Wang, Y., Yu, Z., Jin, M., 2019. E-commerce supply chains under capital constraints. Electronic Commerce Research and
Applications 35, 100851.
Wang, Y.Y., Tao, F., Wang, J., 2021. Information disclosure and blockchain technology adoption strategy for competing
platforms. Information & Management 103506.
Wongkitrungrueng, A., Assarut, N., 2020. The role of live streaming in building consumer trust and engagement with
social commerce sellers. Journal of Business Research 117, 543–556.
Xia, J., Niu, W., 2019. Adding clicks to bricks: an analysis of supplier encroachment under service spillovers. Electronic
Commerce Research and Applications 37, 100876.
Xiao, L., Wang, X.J., Chin, K.S., 2020. Trade-in strategies in retail channel and dual-channel closed-loop supply chain
with remanufacturing. Transportation Research Part E: Logistics and Transportation Review 136, 101898.
Xie, J., Wei, L., Zhu, W., Zhang, W., 2021. Platform supply chain pricing and financing: who benefits from e-commerce
consumer credit? International Journal of Production Economics 242, 108283.
Xu, X., Choi, T.M., 2021. Supply chain operations with online platforms under the cap-and-trade regulation: im-
pacts of using blockchain technology. Transportation Research Part E: Logistics and Transportation Review 155,
102491.
Xu, X., He, P., Fan, Y., 2022. The pricing and carbon abatement decisions of a manufacturer selling with marketplace or
reselling mode. International Transactions in Operational Research 29, 2, 1220–1245.
Xue, M., Zhang, J., Zhu, G., 2020. Quantity decision timing with spillover effect and asymmetric demand information.
Transportation Research Part E: Logistics and Transportation Review 142, 102048.
Yoo, B., Jang, M., 2019. A bibliographic survey of business models, service relationships, and technology in electronic
commerce. Electronic Commerce Research and Applications 33, 100818.
Yoo, S.H., Choi, T.Y., Kim, D., 2021. Integrating sourcing and financing strategies in multi-tier supply chain manage-
ment. International Journal of Production Economics 234, 108039.
Zennyo, Y., 2020. Strategic contracting and hybrid use of agency and wholesale contracts in e-commerce platforms.
European Journal of Operational Research 281, 1, 231–239.
Zhang, C., Li, Y., Ma, Y., 2021a. Direct selling, agent selling, or dual-format selling: electronic channel con-
figuration considering channel competition and platform service. Computers & Industrial Engineering 157,
107368.
Zhang, M., Liu, Y., Wang, Y., Zhao, L., 2022a. How to retain customers: understanding the role of trust in live streaming
commerce with a socio-technical perspective. Computers in Human Behavior 127, 107052.
Zhang, X., Li, Q., Liu, Z., Chang, C.T., 2021b. Optimal pricing and remanufacturing mode in a closed-loop supply chain
of WEEE under government fund policy. Computers & Industrial Engineering 151, 106951.
Zhang, Z., Xu, H., Ke, G.Y., Chen, K., 2022b. Selecting online distribution modes for differentiated products in a plat-
form supply chain. International Journal of Production Economics 244, 108384.
Zhao, D., Han, H., Shang, J., Hao, J., 2020. Decisions and coordination in a capacity sharing supply chain under fixed
and quality-based transaction fee strategies. Computers & Industrial Engineering 150, 106841.
Zhao, D., Yuan, Z., Chen, M., Yang, S., 2022. Differential pricing strategies of ride-sharing platforms: choosing cus-
tomers or drivers? International Transactions in Operational Research 29, 2, 1089–1131.
Zhou, Q., Meng, C., Yuen, K.F., Sheu, J.B., 2021. Remanufacturing authorization strategy for an original equipment
manufacturer-contract manufacturer supply chain: cooperation or competition? International Journal of Production
Economics 240, 108238.

Appendix

∂ 2 πM
NL
Proof of Theorem 1. In the model NL, since 2 = 2βe (φ − 1) < 0, πM
NL
is concave on pNL
e . It
∂ (pNL
√e NL)
∂πMNL α+ε ke
can be obtained from ∂ pNL
= 0 that pNL
e = 2βe
+ cm
2(1−φ)
. Substituting pNL
e into πENL , we have
e

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1118 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
∂ 2 πENL φαε ∂πENL α2
2
√
=− < 0. πENL is concave on kNL
e . It can be obtained from ∂kNL
= 0 that kNL∗
e = ε2
.
∂ (kNL
e ) 8βe k3e e
2α(1−φ)+βe cm
Substituting kNL∗
e into pNL NL∗
e , we have pe = 2βe (1−φ)
. 

Proof of Theorem 2. In the model ML, the Hessian matrix of πM ML

in terms of pMLe and pML
l
−
is H ML1 = [ 0
2βe (φ 1)
2βl (λ − 1)]. From the assumptions 0 < φ < 1 and 0 < λ < 1, we have |H1
0 ML1
|=
2βe (φ − 1) < 0 and |H2ML1 | = 4βe βl (1 − φ)(1 − λ) > 0. Hence, πM ML
is strictly
√ concave with respect
∂π ML ∂π ML α+ρsML +ε kML
to pML
e and pMLl . From ∂ pML
M
= 0 and ∂ pMML = 0, then pML
e = 2βe
e
+ 2(1−φ)
cm
and pMLl =
√ e l
α+sML +ε kML
2βl
l
+ 2(1−λ)
cm
.
By substituting pe and pML
ML
l into πLML , the Hessian matrix of πLML in terms of kML
l and sML is
ελ(α+s) ελ
√3 √
H ML2 = [ 8βlελ kl 4βλ
l kl
]. From the assumptions βl ≥ λ2 and 0 < λ < 1, we have |H1ML2 | = ελ(α+s)
√ <
√
4β k 2β
− 1 8βl k3l
l ll √
0 and |H2ML2 | = ελ (2βl −λ)(α+s)+ελ
2
√3 kl
> 0. Hence, πLML is strictly concave with respect to kML l and
16βl kl
∂π ML ∂πLML
= αε2 and sML∗ = 2β2λα
2
sML . From ∂kLML = 0 and ∂sML = 0, then kML∗ l l −λ
. Substituting kML∗
l and sML∗
l
∂ 2 πEML φεα(2βl −λ+2ρλ)
into πEML , 2 = −
√ < 0. Therefore, πEML is concave on kML e . It can be obtained
∂ (kMLe ) 8βe (2βl −λ) k3e
∂π ML 2
α 2 (2βl −λ+2ρλ)2 +α(2βl −λ+2ρλ)
= α (2β l −λ+2ρλ)
2
from ∂kEML = 0 that kML∗ e ε (2βl −λ)
2 2 . Then, we have pML∗
e = 2βe (2βl −λ)
+ 2(1−φ)
cm
,
e

plML∗
= 2βl −λ + 2(1−λ) .
2α cm


Proof of Theorem 3. The proof of Theorem 3 is similar to Theorem 2, so we omit it. 

Proof of Proposition 1. From the assumptions βl ≥ λ2 , 0 < φ < 1, and 0 < λ < 1, by examining the
optimal retail prices of E-commerce platform and live streaming channels in the three models NL,
ML, and EL, it is easily verified that

∂ pNL∗ ∂ pML∗ ∂ pEL∗ ∂ pML∗ ∂ pEL∗

(1) e
∂cm
= e
∂cm
= ∂cem = 2(1−φ) 1
> 0, ∂cl m = ∂cl m = 2(1−λ) 1
> 0;
∂ pML∗ ∂ ML∗
∂ pe
EL∗ ∂ pEL∗
= βλεe (2β (1+α(2βl −λ+2ρλ))
= βλε (1−φ)(1+2βl −λ+λφ+2ρλ(1−φ))
2 p 2
(2) e
∂ρ l −λ)(2βl −λ+2ρλ)
> ∂ρ l
= 0, ∂ρ e (2βl −λ+λφ)(2βl −λ+λφ+2ρλ(1−φ))
> ∂ρl = 0.
∂ pML∗ ∂ pEL∗
− ∂ρe = λε ) + λεβeα ( 2βl1−λ
2 2
Since e
∂ρ
( 1
βe (2βl −λ)(2βl −λ+2ρλ)
− (2βl −λ+λφ)(2βl1−φ
−λ+λφ+2ρλ(1−φ))
−
1−φ ∂ pML∗ ∂ pEL∗ ∂ pML∗ ∂ pEL∗
(2βl −λ+λφ)
) > 0, thus, ∂ρ e
> ∂ρe > ∂ρ l
= ∂ρl = 0;
∂ pML∗ ∂ pEL∗
(3) l
∂λ
= (2β2α−λ)2 + 2(1−λ) cm
2 > ∂λ
l
= (2β2(1−φ)α 2 +
cm
2(1−λ)2
> 0. 
l l −λ+λφ)

Proof of Proposition 2. From the assumptions βl ≥ λ2 , 0 < φ < 1, and 0 < λ < 1, by examining the
optimal advertising investments in the three models NL, ML, and EL, it is easily verified that

∂kNL∗ ∂kML∗ l −λ+2ρλ)

2
∂kEL∗ 2
(1) e
∂α
= 2α
ε2
> 0, ∂α e
= 2α(2β
ε 2 (2βl −λ)2
> 2αε2
> 0, ∂α e
= 2α(2βlε−λ+λφ+2ρλ(1−φ))
2 (2β −λ+λφ)2
> 2α
ε2
> 0, since
l
2βl −λ+2ρλ ∂k ML∗
∂k EL∗

2βl −λ
− 2βl −λ+λφ+2ρ(1−φ)
2βl −λ+λφ
= 2ρλ (2βλφ+φ(2β l −λ)
l −λ)(2βl −λ+λφ)
> 0, thus, ∂α e
− ∂α e
> 0;

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1119
∂kML∗ 2
l −λ+2ρλ) ∂kEL∗ 2
l −λ+λφ+2ρλ(1−φ)) ∂kML∗ ∂kEL∗
(2) e
∂ρ
= 4λαε2(2β (2βl −λ)2
> 0, e
∂ρ
= 4λα (1−φ)(2β
ε 2 (2βl −λ+λφ)2
> 0, since e
∂ρ
− ∂ρe =
4λα 2 (1−φ) 2ρλ 2ρλ ∂kML∗ ∂kEL∗
ε2
1
( (2βl −λ)(1−φ) − 2βl −λ+λφ
1
+ (2β −λ) 2
(1−φ)
− (2β −λ+λφ) 2 ) > 0, thus,
e
∂ρ
> ∂ρe > 0. 
l l

Proof of Proposition 3. From the assumptions βl ≥ λ2 , 0 < φ < 1, and 0 < λ < 1, by examining the
optimal service investments and service levels in the two models ML and EL, it is easily verified
that
∂kML∗ ∂kEL∗ ∂kML∗ ∂kEL∗ 2
(1) l
∂α
= ∂α l
= 2α
ε2
> 0, ∂ε l
= ∂εl = − 2α ε3
< 0;
∂sML∗
∂sEL∗ ∂sML∗ ∂sEL∗
(2) ∂α
= 2βl −λ >
2λ
∂α
= 2βl −λ+λφ > 0, ∂λ = (2β4αβ
2λ(1−φ) l
2 > ∂λ
= 4αβl (1−φ)
(2βl −λ+λφ)2
> 0. 
l −λ)

2βl −λ−2ρλ 2βl −λ+λφ−2ρλ(1−φ)

Proof of Proposition 4. Denoting 1 = (2βl −λ+2ρλ)2
, 2 =
(2βl −λ+λφ+2ρλ(1−φ))2
, 3 =
4βl λρφ λ
From the
(2βl −λ)(2βl −λ+λφ+2ρλ(1−φ))2 −(2βl −λ+λφ)(2βl −λ+2ρλ)2
. assumptions βl ≥ 2 , 0 < φ <
1, and
0 < λ < 1, comparing the optimal retail prices of E-commerce platform and live streaming
channels in the three models NL, ML, and EL, we have
(2βl −λ+2ρλ)2 α 2 −(2βl −λ−2ρλ)α
(1) pML∗
e − pNL∗
e = 2βe (2βl −λ)
, if α > 1 , then pML∗
e − pNL∗
e > 0, oth-
2 2
(2βl −λ+λφ+2ρλ(1−φ)) α −(2βl −λ+λφ−2ρλ(1−φ))α
erwise, pe − pe ≤ 0;
ML∗ NL∗
p e − pe =
EL∗ NL∗
2βe (2βl −λ+λφ)
,
if α > 2 , then pe − pe > 0, otherwise,
EL∗ NL∗
pe − pe ≤ 0;
EL∗ NL∗
pe − pe
EL∗ ML∗
=
(2βl −λ)(2βl −λ+λφ+2ρλ(1−φ))2 α 2 −(2βl −λ+λφ)(2βl −λ+2ρλ)2 α 2 −4βl λρφα
2βe (2βl −λ)(2βl −λ+λφ)
, if α > 3 , then pe − pe EL∗ ML∗
> 0,
otherwise, pe − pe
EL∗ ML∗
≤ 0;
2φλα
(2) pl − p l
EL∗ ML∗
= − (2βl −λ)(2βl −λ+λφ) < 0. 

Proof of Proposition 5. From the assumptions βl ≥ λ2 , 0 < φ < 1, and 0 < λ < 1, comparing the
optimal advertising investments, service investments and service levels in the three models NL, ML,
and EL, we have
4α 2 ρλ(2βl −λ+ρλ)
= 4(1−φ)α ρλ(2β l −λ+λφ+ρλ−ρφλ)
2
(1) kML∗
e − kNL∗
e = ε 2 (2βl −λ)2
> 0, kEL∗ e − kNL∗
e ε 2 (2βl −λ+λφ)2
> 0, kEL∗
e −
2 2
= −2ρλφ(2β l −λ)−2ρλ φ
2
= αε2 [(1 + 2β
2 2ρλ(1−φ) 2ρλ(1−φ)
kML∗
e l −λ+λφ
) − (1 + 2β2ρλ
l −λ
) ], since 2β l −λ+λφ
− 2β2ρλ
l −λ (2βl −λ+λφ)(2βl −λ)
< 0, and
thus kEL∗
e − kML∗
e < 0. Therefore, we have kML∗ e > kEL∗e > kNL∗
e ;
(2) ML∗
kl − kl = 0;
EL∗
4αβl λφ
(3) sEL∗ − sML∗ = − (2βl −λ)(2β l −λ+λφ)
< 0. 
2βl −λ+6ρλ 2βl −λ+λφ+6ρλ(1−φ)
Proof of Proposition 6. Denoting 4 =
(2βl −λ+2ρλ)2
, 5 = (2β −λ+λφ+2ρλ(1−φ))2 , 6 =
√ √ l √
(6βl −λ+6ρλ)(1−λ)− (6βl −λ+6ρλ)(1−λ)+ (6βl −λ+λφ+6ρλ(1−φ))(1−λ)−
, 7 = , 8 = , 9 =
1 1 2
2(2βl −λ+2ρλ)2 (1−λ) √ 2(2βl −λ+2ρλ)2 (1−λ) 2(2βl −λ+λφ+2ρλ(1−φ))2 (1−λ)
(6βl −λ+λφ+6ρλ(1−φ))(1−λ)+
2(2βl −λ+λφ+2ρλ(1−φ))2 (1−λ)
2
, 1 = (6βl − λ + 6ρλ)2 (1 − λ)2 − 4βl cm (2βl − λ + 2ρλ)2 (1 − λ)(2βl −
2 2 2
λ) and 2 = (6βl − λ + λφ + 6ρλ(1 − φ)) (1 − λ) − 4βl cm (2βl − λ + λφ + 2ρλ(1 − φ)) (1 −
λ)(2βl − λ + λφ). From the assumptions βl ≥ λ2 , 0 < φ < 1, and 0 < λ < 1, comparing the opti-
mal demands of E-commerce platform and live streaming channels and total demands of EPSC in
the three models NL, ML, and EL, we have

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
1120 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121
α(2βl −λ+6ρλ−α(2βl −λ+2ρλ)2 )
(1) DML∗
e − DNL∗
e = 2(2βl −λ)
, if α < 4 , then DML∗ e − DNL∗e > 0, oth-
2
α(2β −λ+λφ+6ρλ(1−φ)−α(2β −λ+λφ+2ρλ(1−φ)) )
erwise, DML∗
e − DNL∗e ≤ 0; DEL∗
e − DNL∗
e = l l
2(2βl −λ+λφ)
,
if α < 5 , then De − De > 0, otherwise, De − De ≤ 0; De − De
EL∗ NL∗ EL∗ NL∗ EL∗ ML∗
=
−αλφ 2ρ(2βl −λ)+2ρλ+α(1+2ρ )(2(2βl −λ+2ρλ(1−φ))+λφ(1+2ρ ))
2(2βl −λ)(2βl −λ+λφ)
< 0, thus, D EL∗
e − D ML∗
e < 0;
−2αβl λφ
(2) DEL∗
l − D ML∗
l = (2βl −λ)(2βl −λ+λφ)
< 0;
2
= −(2βl −λ+2ρλ) )(1−λ)α2(2β +(6βl −λ+6ρλ)(1−λ)α−βl cm (2βl −λ)
2
(3) DtML∗
− Dt NL∗
l −λ)(1−λ)
, if 6 <α <

7, then DtML∗
− Dt NL∗
> 0, otherwise, Dt ML∗
− Dt NL∗
≤ 0; Dt − DtNL∗ =
EL∗
2
−(2βl −λ+λφ+2ρλ(1−φ)) (1−λ)α +(6βl −λ+λφ+6ρλ(1−φ))(1−λ)α−βl cm (2βl −λ+λφ)
2

2(2βl −λ+λφ)(1−λ)
, if 8 <α <

9, then Dt − Dt
EL∗ NL∗
> 0, otherwise, Dt − Dt
EL∗ NL∗
≤ 0; Dt − DtML∗ =
EL∗
2ρ(2β−λ)+2ρλ+α(1+2ρ )(2(2β−λ+2ρλ(1−φ))+λφ(1+2ρ ))+4β
−αλφ 2(2β−λ)(2β−λ+λφ)
< 0. 

Proof of Propositions 7-9. We calculate the differences of the EPSC members’ profits among three
scenarios, respectively. Then, we obtain the following equations and conditions.
(α 2 γ 2 +αγ )(1−φ) 3αγ −α 2 γ 2 βe cm βl cm
Denoting F1 = ( 3 2βe γ31 − c2m )( 32γ1 3 − 2(1−φ) ) + ( 2α(1−λ)
γ1
− c2m )( 2αβ
γ1
l
− 2(1−λ) )−
(2α(1−φ)−βe cm )2 (α 2 γ 2 +αγ )(1−φ) 3αγ −α 2 γ 2 βe cm βl cm
4βe (1−φ)
, F2 = ( 3 2βe γ31 − c2m )( 32γ1 3 − 2(1−φ) ) + ( 2α(1−λ)
γ1
− c2m )( 2αβ
γ1
l
− 2(1−λ) )−
(α 2 γ42 +αγ4 )(1−φ) 3αγ −α γ
2 2
βe cm βl cm α γ
2 2
+αγ
( 2βe γ2
− c2m )( 42γ2 4 − 2(1−φ) ) − ( 2α(1−λ)
γ2
− c2m )( 2αβ
γ2
l
− 2(1−λ) ), F3 = φ( 2β4 e γ2 4 +
3αγ −α 2 γ 2 βe cm βl cm
2
(4α 2 (1−φ) −βe2 c2m )φ α 2 γ 2 +αγ
cm
2(1−φ)
)( 42γ2 4 − 2(1−φ) ) + λφ( 2α γ2
+ 2(1−λ)
cm
)( 2αβ
γ2
l
− 2(1−λ) )− 4βe (1−φ)2
, F4 = φ( 2β4 e γ2 4 +
3αγ −α 2 γ 2 βe cm βl cm α 2 γ 2 +αγ 3αγ −α 2 γ 2
cm
)( 42γ2 4 − 2(1−φ) ) + λφ( 2α γ2
+ 2(1−λ)
cm
)( 2αβ
γ
l
− 2(1−λ) ) − φ( 2β3 e γ1 3 + 2(1−φ) cm
)( 32γ1 3 −
2(1−φ)  2
βe cm α(1−λ) 16β ρλ(1−λ)(2βl +ρλ(1−φ))−βl γ42 (αγ4 +1)(3−αγ4 )+4βl γ22
), C1 = γ2 β l
[4(β l + ρλ(1 − φ)) − (1 − φ) e ],
2(1−φ)  (1−λ)βe
16β ρλ(1−λ)(2βl +ρλ(1−φ))−βl γ4 (αγ4 +1)(3−αγ4 )+4βl γ2
2 2
C2 = α(1−λ) γ2 β l
[4(βl + ρλ(1 − φ)) + (1 − φ) e (1−λ)βe
], C3 =
(4αβe γ12 −αγ32 (αγ3 +1)(3−αγ3 ))(1−φ)
βe γ1 γ3 ((3−αγ3 )−βe (αγ3 +1))
and C4 = √ 4α(1−λ)
.
(2βl −λ)(2βl −λ+λφ)

(1) By comparing M’s optimal profits among the three scenarios, we can obtain: πM
ML∗
− πM
NL∗
=
(α 2 γ 2 +αγ )(1−φ) 3αγ3 −α 2 γ32 βe cm βl cm 2
( 3 2βe γ31 − c2m )( 2γ1
− 2(1−φ) ) + ( 2α(1−λ)
γ1
− c2m )( 2αβ
γ1
l
− 2(1−λ) ) − (2α(1−φ)−β
4βe (1−φ)
e cm )
− Fl ,
when Fl < F1 , then πM − πM > 0, otherwise πM − πM ≤ 0; πM − πM
ML∗ NL∗ ML∗ NL∗ EL∗ NL∗
=
2 (αγ4 +γ4 )(1−φ)(3γ4 −αγ4 )+16βe βl (1−λ)−4(1−φ)γ2
2 2 2
βl βl +ρλ(1−φ)
c − 2α
4(1−λ) m
2
γ2
cm + α 4βe γ22
, when 0 < cm < C1
and cm > C2 , then πM EL∗
− πM NL∗
> 0, otherwise πM EL∗
− πM NL∗
≤ 0; πEML∗ − πENL∗ =
(3γ3 −αγ32 )−(αγ32 +γ3 )βe αγ32 (αγ3 +1)(3−αγ3 )−4αβe γ12
αφ( 4βe (1−φ)γ1
cm + 4βe2 γ12
), when cm > C3 , then πEML∗ − πENL∗ > 0,
otherwise πEML∗ − πENL∗ ≤ 0;
(2) By comparing E’s optimal profits among the three scenarios, we can obtain: πEML∗ − πENL∗ =
(3γ −αγ 2 )−(αγ 2 +γ )β αγ 2 (αγ +1)(3−αγ )−4αβe γ12
αφ( 3 4β3e (1−φ)γ3 1 3 e cm + 3 3 4β 2 γ 2 3 ), when cm > C3 , then πEML∗ − πENL∗ > 0, oth-
e 1
α γ +αγ
2 2
3αγ −α 2 γ 2 βe cm
erwise πEML∗ − πENL∗ ≤ 0; πEEL∗ − πENL∗ = φ( 2β4 e γ2 4 + 2(1−φ) cm
)( 42γ2 4 − 2(1−φ) ) + λφ( 2α γ2
+
2
2αβl βl cm (4α (1−φ) −βe cm )φ
2 2 2
cm
2(1−λ)
)( γ2 − 2(1−λ) ) − 4βe (1−φ)2
− Fl , when Fl < F3 , then πEEL∗ − πENL∗ > 0, other-

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.
 X. Zhang et al. / Intl. Trans. in Op. Res. 31 (2024) 1093–1121 1121
α 2 γ42 +αγ4 3αγ −α 2 γ 2 βe cm
wise πEEL∗ − πENL∗ ≤ 0; πEEL∗ − πEML∗ = φ( 2βe γ2
+ 2(1−φ)
cm
)( 42γ2 4 − 2(1−φ) ) + λφ( 2α
γ2
+
βl cm α γ
2 2
+αγ 3αγ −α γ
2 2
βe cm
cm
2(1−λ)
)( 2αβ
γ2
l
− 2(1−λ) ) − φ( 2β3 e γ1 3 + 2(1−φ)
cm
)( 32γ1 3 − 2(1−φ) ) − Fl , when Fl < F4 , then
πE − π E
EL∗ ML∗
> 0, otherwise πE − πEEL∗ ML∗
≤ 0;
(3) By comparing L’s optimal profits among the three scenarios, we can obtain: πLEL∗ − πLML∗ =
c2 (2βl −λ)(2βl −λ+λφ)−16α 2 (1−λ)2
βl λφ m 4(1−λ) 2
(2βl −λ)(2βl −λ+λφ)
, when cm > C4 , then πLEL∗ − πLML∗ > 0, otherwise πLEL∗ −
πLML∗ ≤ 0. 

© 2022 The Authors.

International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.