Accepted Manuscript

Pricing decisions for substitutable products with green manufacturing in a competitive

supply chain

Peng Ma, Chen Zhang, Xianpei Hong, Henry Xu

PII: S0959-6526(18)30465-7
DOI: 10.1016/j.jclepro.2018.02.152
Reference: JCLP 12099

To appear in: Journal of Cleaner Production

Received Date: 8 April 2017

Revised Date: 5 January 2018
Accepted Date: 14 February 2018

Please cite this article as: Ma P, Zhang C, Hong X, Xu H, Pricing decisions for substitutable products
with green manufacturing in a competitive supply chain, Journal of Cleaner Production (2018), doi:
10.1016/j.jclepro.2018.02.152.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
Pricing decisions for substitutable products with green
manufacturing in a competitive supply chain

Peng Maa, Chen Zhanga, Xianpei Hongb,*, Henry Xuc

a
School of Management Science and Engineering, Nanjing University of Information Science
and Technology, Nanjing 210044, China
b
College of Economics and Management, Huazhong Agricultural University, Wuhan 430070,

PT
China
c
UQ Business School, The University of Queensland, Brisbane, Australia 4072

RI
Abstract: In this paper, we examine the pricing strategies in a two-stage supply chain with
two competitive manufacturers and one retailer. We address six game models: the centralized

SC
model (Model I), the MS-Bertrand model (Model II), the MS-Stackelberg model (Model III),
the RS-Bertrand model (Model IV), the RS-Stackelberg model (Model V) and the
cost-sharing contract model (Model VI) to explore the optimal pricing strategies of

U
substitutable products. We address the optimal green manufacturing level, retail prices,
AN
wholesale prices and the profits of supply chain members as well as the whole supply chain
under different models. Numerical examples are provided to demonstrate the efficiency and
effectiveness of the proposed models. First the impact of green investment on the green
M

manufacturing level and supply chain performance is examined. Then the impact of price
elasticity, cross-price sensitivity and green manufacturing coefficient on the green
D

manufacturing level is analyzed. We find that the centralized model is the best, and the
cost-sharing contract model will be better than the four decentralized models when the
TE

cost-sharing proposition is in a certain interval. Additionally, in decentralized scenarios, the

Stackelberg model has an advantage for manufacturers while the Bertrand model is superior
EP

for the retailer. Our results also indicate that green manufacturing will benefit the
manufacturer involved in green investment.
Keywords: Green supply chain; Pricing strategies; Green investment; Cost-sharing contract;
C

Green manufacturing level

AC

1. Introduction
With the development of society and the economy and the increased standard of living,
green manufacturing gradually gains awareness. Consumers and companies are more
conscious of environmental issues and the risks posed by unmitigated climate change
(Fahimnia et al., 2015; Laari et al., 2016). Today, increasing numbers of consumers buy
no-pollution and environmentally harmless green products, while many manufacturers and
retailers produce or sell green products to enhance their competitive advantage (Li et al., 2015;
Ülkü et al., 2017). Consumers prefer to pay more for low-carbon, energy saving and
1
 ACCEPTED MANUSCRIPT
eco-protective products, driving green industry development. Moreover, national
manufacturing strategies such as “German industrial 4.0” and “Made in China 2025” also
contribute to the development of green manufacturing. In this study, we are interested in
analyzing the impact of green investment on green manufacturing level and supply chain
performance, as well as the impact of price elasticity, cross-price sensitivity and green
manufacturing coefficient on the green manufacturing level. We aim to examine how the
supply chain members maximize their profits in a competitive environment.

PT
The negative effects of climate change caused by industrial activities are inevitable.
Therefore, countries and businesses around the world hope to find more eco-friendly

RI
industrial management systems or develop efficient supply chain management (Sulistio and
Rini, 2015; Kumara et al., 2015). In addition, given the influence of complicated markets and

SC
multifaceted challenges, recent studies show that sustainable supply chain management
(SSCM) has become ever more important and thus has received increased attention from
academics and practitioners (Bechtsis et al., 2017; Ansari and Kant, 2017). To an extent,

U
SSCM is an extension of green supply chain management (GSCM). The goal of a green
AN
supply chain is to improve supply and distribution efficiencies, save time and money and
satisfy customer needs (Bhattacharya et al., 2015). GSCM efficiently improves brand image,
then captures an increased market share and gains customer trust by providing desired and
M

quality products (Sharma et al., 2017). Therefore, GSCM can contribute to sustainable
performance enhancement, leading to SSCM performance and green innovation (Chin et al.,
D

2015). To improve green performance, GSCM is crucial and must be addressed (Jabbour et al.,
2015; Geng et al., 2017).
TE

In the practices of manufacturing corporations, GSCM is closely related to many

activities, including green design, green operations, green innovation, green manufacturing,
EP

etc. (Seman et al., 2012). However, manufacturing will not develop without government
financial intervention (Hafezalkotob, 2015 and 2017). Moreover, GSCM is influenced by
internal and external factors, while industry, people, and government are regarded as three
C

vital sectors for GSCM (Ma et al., 2014; Kuei et al., 2015; Li et al., 2015; Zhang et al., 2016).
AC

Many enterprises choose GSCM to be a vital part of their strategic plan. Additionally,

multiple international enterprises stipulate that their suppliers produce greener products.

Therefore, their suppliers must reduce resource consumption and improve production

efficiency during the manufacturing process. Consequently, a green management strategy can

not only create a new market opportunity but also improve corporate reputation awareness,

which increases the competitiveness of the enterprise. Although supply chain members differ

in bargaining power, they are all in pursuit of maximum profits and minimum risk of potential
2
 ACCEPTED MANUSCRIPT

losses (Laari et al. 2017). Many authors have investigated the GSCM from different

perspectives (e.g., Ma et al., 2014; Kuei et al., 2015; Li et al., 2015; Zhang et al., 2016; Laari

et al. 2017). However, it remains unclear how supply chain members make pricing strategies

in a competitive, green-manufacturing context, and how the green investment of one

manufacturer affects the supply chain performance.

PT
To address the above problems, we study a two-stage supply chain composed of two

competitive manufacturers and one retailer. We discuss the optimal results under six models:

RI
the centralized model (Model I), the decentralized model involving MS-Bertrand model

(Model II), MS-Stackelberg model (III), RS-Bertrand model (IV) and RS-Stackelberg model

SC
(Model IV), as well as the cost-sharing contract model (Model VI) based on the MS-Bertrand

model. We also analyze the effects of green investment on the green manufacturing level and

U
on the supply chain performance. In addition, we analyze the impacts of the price elasticity,
AN
the cross-price sensitivity, and the consumer sensitivity on the green manufacturing level.
Our contributions are as follows. First, our paper focuses on the pricing decisions of the
green supply chain in different scenarios. The results will enhance and expand the GSCM
M

research. Second, we consider the competition between two manufacturers and introduce the
manufacturer’s green strategy into the model. Most previous studies concentrate on a green
D

supply chain consisting of one manufacturer and one or two retailers. Different from them, we
address a horizontal competition of Manufacturer 1 (M1) and Manufacturer 2 (M2) in the
TE

supply chain and explore how the green investment affects the supply chain performance. The
difference between the two manufacturers is that M1 is the green manufacturer and M2 is the
EP

traditional manufacturer. We also assume that the two products are substitutable. Finally,
under the cost-sharing contract model, we assume that M1 adopts green manufacturing during
the entire manufacturing process and the retailer accepts the terms offered in the contract.
C

Moreover, we focus on the influence of the green investment parameter, price elasticity,
AC

cross-price sensitivity and the green manufacturing coefficient on the green manufacturing
level and the supply chain performance.
The remainder of this paper is organized as follows. We review related literature in
Section 2. In Section 3, we describe the notations and assumptions of the model. Sections 4
and 5 analyze the centralized and decentralized models respectively. The cost-sharing contract
model is analyzed in Section 6. Section 7 presents the numerical comparison. Finally, we
provide conclusions in Section 8.

3
 ACCEPTED MANUSCRIPT
2. Related literature
Our study encompasses three research streams: optimal pricing strategies of substitutable
products in supply chains, optimal pricing strategies under the green supply chains, and
cost-sharing and revenue-sharing contracts in supply chains.
2.1. Optimal pricing strategies of substitutable products in supply chains
Customers may not always be able to find a preferred product, which results in unmet

PT
demand that is often substituted with an alternative (Shin et al., 2015). Therefore, pricing
strategies of substitutable products have become a popular research area in the supply chain
management. Hsieh and Wu (2009) find that supply chain profits in all models increase with a

RI
higher degree of product substitutability. Gürler and Yılmaz (2010) address a single-period,
newsboy-type inventory problem with two substitutable products. Bish and Suwandechochai

SC
(2010) study the degree of substitution between products and the level of operational
postponement for the optimal capacity and the resulting expected profit. Zhao et al. (2012)
study the pricing problem of substitutable products with one manufacturer and two

U
competitive retailers in a fuzzy environment. Zhao et al. (2014) analyze the pricing decisions
AN
of two competitive manufacturers and one common retailer for two substitutable products.
Zhang et al. (2015) consider a one-manufacturer, one-retailer supply chain and investigate
two types of substitutable products: environmental and traditional. Şen (2016) studies a
M

duopoly model consisting of two firms selling their fixed stocks of two substitutable items
over a selling season. Different from them, we focus on pricing strategies and studying the
D

effects of green investment on supply chain performance.

TE

2.2. Optimal pricing strategies under green supply chains

A reasonable pricing strategy is critical for green products to acquire market share. Tang
and Yin (2007) address a base model with deterministic demand to examine how a retailer
EP

determines both the order quantity and the retail prices of two substitutable products under
fixed and variable pricing strategies. Chen and Sheu (2009) demonstrate that the appropriate
C

design of environmental-regulation pricing strategies promotes extended product

responsibility for firms in a competitive market. Karakul and Chan (2010) study the joint
AC

pricing and procurement of two, one-way substitutable products. Jia and Zhang (2013)
investigate the dynamic pricing and ordering decisions for a durable product with multiple
generations in the context of one manufacturer and one retailer. Chen et al. (2013) study the
pricing policies in a supply chain with one manufacturer selling a product to an independent
retailer and directly to consumers through an Internet channel. Ceryan et al. (2013) consider a
firm producing substitutable products via a capacity portfolio consisting of both
product-dedicated and flexible resources and characterize the structure of the optimal
production and pricing decisions. Li et al. (2016) introduce e-commerce into GSCM and

4
 ACCEPTED MANUSCRIPT
discuss the pricing and green strategies in both centralized and decentralized models. Bajwa
et al. (2016) address a multi-product pricing and production problem in a discrete-time setting
with capacity constraints and setup costs. Under the Big Data environment, Liu and Yi (2017)
propose four game situations and analyze the pricing policies with targeted advertising and
products green degree. By contrast, our research focuses on the pricing strategies of
substitutable products in a supply chain where only one manufacturer invests in green
manufacturing.

PT
2.3. Cost-sharing and revenue-sharing contracts in supply chains
The formulation and implementation of an effective coordination strategy among

RI
member companies is key to the promotion of sustainable development of a green supply
chain (Cao and Zhang, 2013). Tsao and Sheen (2012) show that an appropriate range for the

SC
fraction of promotion cost sharing can achieve channel coordination and ensure the
distribution of increasable profit. Analogously, cost and revenue sharing can be a measure of
channel coordination (Kunter, 2012; Ma et al., 2013). Ghosh and Shah (2015) explore supply

U
chain coordination problems arising from green initiatives and explore the impact of a
AN
cost-sharing contract on the decisions of supply chain members undertaking green initiatives.
With cost-sharing and wholesale price premium contracts, the retailer can achieve the goal of
reducing carbon emissions jointly with the manufacturer, which can encourage the
M

manufacturer to increase its carbon emission reduction rate and enhance the profits of the
supply chain (Wang et al., 2016). Lastly, Dai et al. (2017) find that a cost-sharing contract
D

brings more profits to supply chain members and the whole supply chain than a
non-cooperative model does. We study how the manufacturer encourages the retailer to sell
TE

green products through sharing of costs in green investment.

3. Model description
EP

We consider a two-stage supply chain with two manufacturers and one retailer. Each
manufacturer sells one product to the retailer and the two products are substitutable to each
C

other. Specially, M1 is a green manufacturer and produces product 1 with green design,
AC

sustainable material and smart technology in the manufacturing process. In contrast, M2 is a

traditional manufacturer and produces product 2 by using conventional design, high energy
consumption and traditional technology. Generally, product 1 use less carbon and thus is
much cleaner, greener and more environmentally friendly than product 2.
To promote industrial transformation, the manufacturers (i.e., M1 and M2) produce
their respective products at different unit production costs ci ( i = 1, 2 ) and sell them at

different unit wholesale prices ωi ( i = 1, 2 ) to the same retailer. The retailer sells the two

products at a respective unit retail price pi ( i = 1, 2 ). Our model setting is shown in Fig. 1.

5
 ACCEPTED MANUSCRIPT

c1

Manufacturer 1 ω1 p1
(M1) green

Retailer Customers
c2 (R) (C)

PT
Manufacturer 2
(M2) original
ω2 p2

RI
Fig.1. Supply chain model structure
We assume that market demand D1 for M1’s products is a linear function of two retail

SC
prices and the green manufacturing level ( θ ), while market demand D2 for M2’s products

is a linear function of the retail prices p1 and p2 . The demand functions are given as

U
follows:
D1 = ε a − α p1 + γ p2 + βθ , (1)
AN
D2 = (1 − ε )a − α p2 + γ p1 . (2)
Where a represents the initial market potential and ε is the degree of customer loyalty to
M

product 1, and 1−ε is the degree of customer loyalty to the product 2; 0 ≤ ε ≤ 1 . The
parameters α and γ measure the price elasticity of demand to its own price and the cross-price
D

sensitivity, respectively. Assume the parameters α and γ satisfy α>γ>0, which means that the
demand for a product is more sensitive to changes in its own price than changes in the price
TE

of the competing product. Parameter β measures consumer sensitivity to the green

manufacturing level (i.e., green manufacturing coefficient). Market demand can be satisfied in
EP

the selling period.

In the demand function, ε is a constant. M1 and M2 divide the market and product
demands are affected by the two prices. Notice that Eq. (1) reflects a “green” sensitive
C

consumer market and it is increasingly linear with the green manufacturing level. We capture
AC

the impact of retail prices and the green manufacturing level on market demand in a tractable
deterministic linear form. Assume that green manufacturing does not influence the marginal
cost of M1’s production. We investigate the effects of green manufacturing level on the
demand for product 1. Additionally, we aim to determine how the green manufacturing level
affects the profits of the manufacturers and the retailer. To model the cost of green
manufacturing, we consider an increasing and convex cost structure that reflects how the
green manufacturing level results in the initial changes in products and processes. The cost of

6
 ACCEPTED MANUSCRIPT
green manufacturing is given by I θ 2 , where I is the green investment parameter (Ghosh and
Shah, 2015). Lastly, in our base model, the green investment is decided by M1.
The basic notations are shown in Table 1 and proofs of Propositions are given in the
Appendix.
Table 1. Notations
Given parameters
a The initial market potential

PT
ߝ The degree of customer loyalty to the product 1
ci Unit manufacturing cost of product 1 (i=1) and product 2 (i=2)

RI
Di The market demand of product 1 (i=1) and product 2 (i=2)
ߙ The price elasticity of demand to its own price

SC
γ The cross-price sensitivity
ߚ The green manufacturing coefficient
߶ The fraction of the green manufacturing cost that M1 pays

U
I The green investment parameter
AN
મ Profit function
Decision variables
ωi The wholesale price of product 1 (i=1) and product 2 (i=2)
M

pi The retail price of product 1 (i=1) and product 2 (i=2)

ߠ The green manufacturing level
D

Superscripts/Subscripts
TE

()* Optimal results

()I,()II,()III Results from the centralized model, the MS-Bertrand model, and the
MS-Stackelberg model, respectively
EP

IV V VI
() ,() ,() Results from the RS-Bertrand model, the RS-Stackelberg model, and the
cost-sharing contract model, respectively
C

()Mi,()R,()SC Results from manufacturer i (i=1, 2), retailer and supply chain, respectively
AC

4. The Centralized model (Model I)

In this section, we consider that only one decision maker determines the retail prices and
the green manufacturing level in the supply chain.
The objective function of the supply chain is as follows:
Π SC = ( p1 − c1 ) D1 + ( p2 − c2 ) D2 − Iθ 2

= ( p1 − c1 )(ε a − ap1 + γ p2 + βθ ) + ( p2 − c2 )[ (1 − ε )a − ap2 + γ p1 ] − Iθ 2 . (3)

The following proposition summarizes the optimal solutions in a centralized model.

I*
Proposition 1. In the centralized model, the optimal retail price of product 1 ( p1 ), the
7
 ACCEPTED MANUSCRIPT
optimal retail price of product 2 ( p2 ), the optimal green manufacturing level ( θ I * ), and the
I*

optimal profit of supply chain ( ΠSC ) are respectively:

I*

2 IA2 − c1αβ 2
p1I * = , (4)
4 IA1 − αβ 2

4 IA3 − A4 β 2
p2I * = , (5)
2(4 IA1 − αβ 2 )

PT
A5 β
θ I* = , (6)
4IA1 − αβ 2

RI
4 IA6 − ( A7 β )2
Π SC
I*
= . (7)
4(4 IA1 − αβ 2 )

SC
A1 –A7 are defined in the Appendix.
Proof. See the Appendix.
Now, we study the impact of the green manufacturing coefficient (β) on the green

U
manufacturing level ߠ.
AN
∂θ I *
Proposition 2. The green manufacturing level ߠ increases with β, i.e., ≥ 0.
∂β
Proof. See the Appendix.
M

Proposition 2 shows that the green manufacturing level ߠ increases with the green
manufacturing coefficient β. This is because that when β becomes lager, its effect on the
D

market demand will increase, then the manufacturer is likely to enhance the green
TE

manufacturing level.
In the following sections, we consider the decentralized model and then discuss the
cost-sharing contract. Lastly, we compare the above-mentioned six models and examine the
EP

effects of model parameters.

5. The Decentralized model

C

5.1. Decision models in MS game

AC

We assume that the MS decentralized decision consists of two larger manufacturers and
one relatively smaller retailer, and the manufacturers control the market. The two
manufacturers and the retailer take a sequential non-cooperative game where two
manufacturers are leaders and the retailer is a follower.
The objective functions of M1, M2 and the retailer are as follows:
ΠM1 = (ω1 − c 1 )(ε a − α p1 + γ p2 + βθ ) − Iθ 2 , (8)

ΠM2 = (ω2 − c2 )[(1 − ε ) a − α p2 + γ p1 ] , (9)

8
 ACCEPTED MANUSCRIPT
Π R = ( p1 − ω1 ) D1 + ( p2 − ω2 ) D2

= ( p1 − ω1 )(ε a − ap1 + γ p2 + βθ ) + ( p2 − ω2 )[(1 − ε )a − ap2 + γ p1 ] . (10)

5.1.1. The MS-Bertrand model (Model II)

Assume that M1 and M2 are the Stackelberg leaders and the retailer is the follower. The
retailer independently determines the retail prices. Based on the retailer’s decisions, M1
determines its wholesale price and the green manufacturing level and M2 sets its wholesale

PT
price simultaneously. The definition of the MS-Bertrand model is given as:
max Π M1 (ω1 , p1* (ω1 , ω2 ,θ ), p2* (ω1 , ω2 ,θ ),θ )
 ω1 ,θ


RI
 max Π M 2 (ω2 , p1* (ω1 , ω2 ,θ ), p2* (ω1 , ω2 ,θ ))
ω2

p1* (ω1 ,ω2 ,θ ), p2* (ω1 , ω2 ,θ )

SC
max Π R ( p1 , p2 )
( p1 , p2 )

The following proposition summarizes the optimal solutions in the MS-Bertrand model.

U
Proposition 3. The optimal values of the wholesale prices ߱ଵூூ∗ and ߱ଶூூ∗ , the green
manufacturing level θII*, the retail prices ‫݌‬ଵூூ∗ and ‫݌‬ଶூூ∗, and profits of M1, M2, the retailer
AN
and the supply chain are as follows:
2 IB2 − c1αβ 2
ω1II * = , (11)
M

IB1 − αβ 2

4 IB3 − A4 β 2
ω2II * = , (12)
2( IB1 − αβ 2 )
D

B4 β
TE

θ II * = , (13)
2( IB1 − αβ 2 )
4 IB5 − B6 β 2
p1II * = , (14)
EP

4 A1 ( IB1 − αβ 2 )

4 IB7 − B8 β 2
p2II * = , (15)
4 A1 ( IB1 − αβ 2 )
C

I (8Iα − β 2 ) B42
Π IIM*1 =
AC

, (16)
4( IB1 − αβ 2 ) 2

α (4 IB9 − A7 β 2 )2
Π MII *2 = , (17)
8( IB1 − αβ 2 )2

[4 IB4 (4 IB10 − A7γβ 2 ) + (4 IB9 − A7 β 2 )(4 IB11 − A7αβ 2 )]α 2

Π IIR * = , (18)
16 A1 ( IB1 − αβ 2 )2

[4IB4 (4IB10 − A7γβ 2 ) + (4IB9 − A7 β 2 )(4IB11 − A7αβ 2 )]α 2

+ A1[4I (8Iα − β 2 ) B42 + 2α (4IB9 − A7 β 2 )2 ]
Π IISC* = . (19)
16 A1 ( IB1 − αβ 2 )2

9
 ACCEPTED MANUSCRIPT
where B1-B11 are defined in the Appendix.
Proof. See the Appendix.
We now study the impact of the green manufacturing coefficient (β) on the green
manufacturing level and obtain Proposition 4.
∂θ II *
Proposition 4. The green manufacturing level increases with β, i.e., ≥ 0.
∂β
Proof. See the Appendix.

PT
Proposition 4 shows that the green manufacturing level ߠ always increases with the
green manufacturing coefficient β. Here, we have the same managerial insights as Proposition

RI
2.

5.1.2. The MS-Stackelberg model (Model III)

SC
When two manufacturers make sequence decisions, we assume that M1 acts as the leader
and M2 acts as a follower. M1 adopts green manufacturing and first announces the wholesale

U
price of product 1 and the green manufacturing level. Then, M2 determines the wholesale
AN
price of product 2 in order to maximize its profit. Finally, based on the decisions made by M1
and M2, the retailer determines the retail prices of the two products. We can formulate the
MS-Stackelberg model as follows:
M

max Π M1 (ω1 , p1* (ω1 , ω2* (ω1 ,θ ),θ ), p2* (ω1 , ω2* (ω1 ,θ ),θ ),θ )
ω1 ,θ

ω2* (ω1 ,θ )
D

max Π M 2 (ω2 , p1* (ω1 , ω2 ,θ ), p2* (ω1 , ω2 ,θ ))

 ω2

TE

 p1* (ω1 , ω2 ,θ ), p2* (ω1 , ω2 ,θ )

 max Π R ( p1 , p2 )
 p1 , p2
EP

The following proposition summarizes the optimal solutions in the MS-Stackelberg

model.
Proposition 5. The optimal values of the wholesale prices ߱ଵூூூ∗ and ߱ଶூூூ∗ , the green
C

manufacturing level θIII*, the retail prices ‫݌‬ଵூூூ∗ and ‫݌‬ଶூூூ∗ , and profits of M1, M2, the retailer
AC

and the supply chain are as follows:

2 IB13 − c1αβ 2
ω1III * = , (20)
IB12 − αβ 2

2 IB14 − A4αβ 2
ω2III * = , (21)
2α ( IB12 − αβ 2 )

B4 β
θ III * = , (22)
2( IB12 − αβ 2 )

10
 ACCEPTED MANUSCRIPT
4 IB15 − B6 β 2
p1III * = , (23)
4 A1 ( IB12 − αβ 2 )

2 IB16 − B17 β 2
p2III * = , (24)
2α A1 ( IB12 − αβ 2 )

IB42
Π MIII1* = , (25)
4α ( IB12 − αβ 2 )

(2 IB18 − A7αβ 2 ) 2

PT
Π MIII2* = , (26)
8α ( IB12 − αβ 2 ) 2

4 B19 I 2 + 4 B20αβ 2 I + A72α 4 β 4

Π III =

RI
*
, (27)
16α A1 ( IB12 − αβ 2 ) 2
R

ΠSC
III *
= ΠMIII1* + ΠMIII2* + ΠIII
R .
*
(28)

SC
where B12-B20 are defined in the Appendix.
Proof. See the Appendix.

U
We now study the impact of the green manufacturing coefficient (β) on the green
AN
manufacturing level and obtain Proposition 6.
∂θ III *
Proposition 6. The green manufacturing level ߠ increases with β, i.e., ≥ 0.
∂β
M

Proof. See the Appendix.

5.2. Decision models in RS game

D

We assume that RS decentralized decisions appear in a market controlled by a powerful

TE

retailer. Assume the retailer is the leader and the two manufacturers are followers in this

scenario. We also use the sequential non-cooperative game to solve the above problem.
EP

5.2.1. The RS-Bertrand model (Model IV)

In this model, the retailer first determines the retail prices of two substitutable products.
C

Having perceived the retail prices, the two manufacturers make their decisions simultaneously.
AC

That is M1 determines the wholesale price of product 1 and the green manufacturing level,

and M2 determines the wholesale price of product 2. We express the RS-Bertrand model as:
max Π R ( p1 , p2 , ω1* ( p1 , p2 ), ω2* ( p1 , p2 ),θ * ( p1 , p2 ))
( p1 , p2 )

ω1* ( p1 , p2 ), ω2* ( p1 , p2 ),θ * ( p1 , p2 )

max Π M (ω1 ,θ )
 ω1 ,θ 1


 max Π M 2 (ω2 )
ω2

The following proposition summarizes the optimal solutions in the RS-Bertrand model.

11
 ACCEPTED MANUSCRIPT
Proposition 7. The optimal values of the wholesale prices ߱ଵூ௏∗ and ߱ଶூ௏∗ , the green
manufacturing level θIV*, the retail prices ‫݌‬ଵூ௏∗ and ‫݌‬ଶூ௏∗ , and profits of M1, M2, the retailer
and the supply chain are as follows:
8C7 I 2 + 2C8 β 2 I − c1γ 2 β 4
ω1IV * = , (29)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

8C9 I 2 + 2C10 β 2 I − c2γ 2 β 4

ω2IV * = , (30)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

PT
(4C11 I + C12γβ 2 ) β
θ IV * = , (31)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

RI
8C13 I 2 + 2C14 β 2 I + C15
p1IV * = , (32)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

SC
8C16 I 2 + 2C17 β 2 I − c2γ 2 β 4
p2IV * = , (33)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

I (4C11 I + C12γβ 2 ) 2 (4α I − β 2 )

U
Π MIV1* = , (34)
[16(C1 I 2 + C2 β 2 I ) − γ 2 β 4 ]2
AN
4α I 2 (4C18 I + C19 β 2 )2
Π MIV2* = , (35)
[16(C1 I 2 + C2 β 2 I ) − γ 2 β 4 ]2
M

2α I (2C20 I + C21β 2 )
Π IV *
= , (36)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
R
D

ΠSC
IV *
= ΠIV
M1 + ΠM2 + ΠR .
* IV * IV *
(37)
TE

where C1, C2, C7-C21 are defined in the Appendix.

Proof. See the Appendix.
∂θ IV *
Proposition 8. The green manufacturing level ߠ increases with β, i.e., ≥ 0.
EP

∂β
Proof. See the Appendix.
C

5.2.2. The RS-Stackelberg model (Model V)

AC

In this case, the retailer first announces the retail price. Then M1 determines the

wholesale price and the green manufacturing level of product 1 to maximize his profit. Finally,

M2 sets the wholesale price of product 2. The RS-Stackelberg model can be formulated as

follows:

12
 ACCEPTED MANUSCRIPT
max Π R ( p1 , p2 , ω1* ( p1 , p2 ), ω2* ( p1 , p2 ),θ * ( p1 , p2 ))
( p1 , p2 )

ω1* ( p1 , p2 ), ω2* ( p1 , p2 ),θ * ( p1 , p2 )

 max Π M (ω1 ,θ , p1 , p2 )
 ω1 ,θ
1

 ω2* ( p1 , p2 )
 max Π (ω , p , p )
 ω2 M2 2 1 2

The following proposition summarizes the optimal solutions in the RS-Stackelberg

PT
model.
Proposition 9. The optimal values of the wholesale prices ߱ଵ௏∗ and ߱ଶ௏∗ , the green

RI
manufacturing level θV*, the retail prices ‫݌‬ଵ௏∗ and ‫݌‬ଶ௏∗ , and profits of M1, M2, the retailer
and the supply chain are as follows:

SC
4C27 I 2 + 2C28αβ 2 I − c1α 2γ 2 β 4
ω1V * = , (38)
8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4

2C29 I 2 + C30 β 2 I − c2γ 2α 3 β 4

U
ω2V * = , (39)
8α (2C22 I 2 + C23 β 2 I ) − γ 2α 3 β 4
AN
(2C31 I + C8γα 2 β 2 ) β
θ V* = , (40)
8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4

4C32 I 2 + C33αβ 2 I + C15α 2

M

p1V * = , (41)
8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4

2C34 I 2 + C35αβ 2 I − c2γ 2α 3 β 4

D

pV2 * = , (42)
8α (2C22 I 2 + C23 β 2 I ) − γ 2α 3 β 4
TE

I (2C31 I + C12γα 2 β 2 )(4C36 I 2 + 2C37αβ 2 I − C12γα 3 β 4 )

ΠVM*1 = , (43)
α [8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4 ]2

I 2 (2C38 I + C39αβ 2 ) 2
EP

ΠVM*2 = , (44)
α [8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4 ]2

I (C40 I + C41α 2 β 2 )
ΠVR* = , (45)
C

8α (2C22 I 2 + C23 β 2 I ) − γ 2α 3 β 4
AC

ΠVSC* = ΠVM*1 + ΠVM*2 + ΠVR* . (46)

where C22, C23, C27-C41 are defined in the Appendix.

Proof. See the Appendix.
∂θ V *
Proposition 10. The green manufacturing level ߠ will increase with β, i.e., ≥0.
∂β
Proof. See the Appendix.

13
 ACCEPTED MANUSCRIPT
We now examine the cost-sharing contract model and address whether the cost-sharing
contract can improve supply chain performance. Specially, we introduce the cost-sharing
contract into the MS-Bertrand model.

6. The Cost-sharing contract model (Model VI)

In this section, we assume that M1 and the retailer share the cost of green
manufacturing. If the retailer actively shares the cost incurred by green manufacturing, it

PT
promotes the development of the green manufacturing process. Therefore, we assume that M1
offers to share φ proportion of the total cost of green manufacturing and the retailer accepts
the offer, incurring 1 − φ proportion of the green manufacturing cost ( 0 < φ ≤ 1 ) (Ma et al.,

RI
2013; Ghosh and Shah, 2015). Based on the above description, the objective functions of M1

SC
and the retailer are as follows:
ΠM1 =(ω1 − c1 )(ε a − α p1 + γ p2 + βθ ) − φ Iθ 2 , (47)

ΠR = ( p1 − ω1)(ε a −α p1 + γ p2 + βθ ) + ( p2 − ω2 )[(1− ε )a −α p2 + γ p1 ] − (1− φ)Iθ 2 . (48)

U ω1III* and ω2III* , the green

AN
Proposition 11. The optimal values of the wholesale prices
III* III*
manufacturing level θIII*, the retail prices p1 and p2 , and the profits of M1, M2, the
M

retailer and the supply chain are as follows:

2φ IB2 − c1αβ 2
ω1VI * = , (49)
φ IB1 − αβ 2
D

4φ IB3 − A4 β 2
ω2VI * = , (50)
TE

2(φ IB1 − αβ 2 )

B4 β
θ VI * = , (51)
2(φ IB1 − αβ 2 )
EP

4φ IB5 − B6 β 2
p1VI * = , (52)
4 A1 (φ IB1 − αβ 2 )
C

4φ IB7 − B8 β 2
pVI *
= , (53)
4 A1 (φ IB1 − αβ 2 )
2
AC

φ I (8φ I α − β 2 ) B42
ΠVIM1* = , (54)
4(φ IB1 − αβ 2 ) 2

α (4φ IB9 − A7 β 2 ) 2
ΠVIM *2 = , (55)
8(φ IB1 − αβ 2 ) 2

4φ Iα 2 B4 (4φ IB10 − A7γβ 2 ) − 4(1 − φ ) IA1 B42 β 2

+α 2 (4φ IB9 − A7 β 2 )(4φ IB11 − A7αβ 2 )
ΠVIR * = , (56)
16 A1 (φ IB1 − αβ 2 )2

14
 ACCEPTED MANUSCRIPT
4φ I (8φ Iα − β 2 ) A1 B42 + 2α (4φ IB9 − A7 β 2 )2 A1 − 4(1 − φ ) IA1 B42 β 2
+4α 2φ I (4φ IB10 − A7γβ 2 ) B4 + α 2 (4φ IB9 − A7 β 2 )(4φ IB11 − A7αβ 2 )
ΠVISC* = . (57)
16 A1 (φ IB1 − αβ 2 )2
where A1, A4, A7, B1-B11 are defined in the Appendix.
Proof. See the Appendix.
We now study the impact of the green manufacturing coefficient (β) on the green
manufacturing level and obtain Proposition 12.

PT
∂θ VI *
Proposition 12. The green manufacturing level ߠ increases with β, i.e., ≥ 0.
∂β

RI
Proof. See the Appendix.
Proposition 12 shows that the green manufacturing level ߠ increases with the green

SC
manufacturing coefficient β. Here, we have the same managerial insights as Propositions 2
and 4. In the next section, we conduct numerical experiments to obtain more insights from
above, and study the effects of the green investment parameter I on the green manufacturing

U
level and supply chain performance, and then study the impacts of parameters α, γ and β on
AN
the green manufacturing level. Lastly, we investigate the effects of cost-sharing proportion (߶)
on the supply chain performance.
M

7. Numerical analysis
In this section, we first study the impact of green investment on the green manufacturing
D

level and supply chain performance. Then, we study the effects of parameter α, γ and β on the
green manufacturing level.
TE

7.1. Comparison and analysis of the optimal values numerically

For the ease of analysis, we provide a numerical example to compare the optimal values
EP

of the wholesale prices, the green manufacturing level, the retail prices and the maximum
profits. The default values of the model parameters are given as follows: a=800, ε=0.6, c1=16,
c2=15, α=6, γ=4, β=3, ϕ=0.6. We let I=3, and then the corresponding results are shown as in
C

Table 2 and Table 3.

AC

Table 2. Optimal values of the wholesale prices, the green manufacturing level and the retail
prices under different models.

Decision scenario p1* ω1* θ* p2* ω2*

Centralized model 139.871 - 61.935 122.081 -

MS-Bertrand model 145.476 70.665 13.664 128.958 57.718

MS-Stackelberg model 150.677 79.146 15.786 131.010 60.549

15
 ACCEPTED MANUSCRIPT

RS-Bertrand model 141.174 48.109 16.054 127.883 34.566

RS-Stackelberg model 146.444 54.464 19.232 130.116 35.846

Cost-sharing contract model 151.571 73.558 23.982 132.538 58.686

PT
RI
Table 3. Optimal profits of supply chain, M1, M2, and the retailer under different models.

SC
Decision scenario Π*SC Π*M1 Π*M2 Π*R

Centralized model 43240.210 - - -

MS-Bertrand model 35274.007

U 8401.555 5474.607 21397.915
AN
MS-Stackelberg model 34947.892 8556.233 6224.009 20167.650

RS-Bertrand model 36593.957 5412.560 2297.066 28884.331

M

RS-Stackelberg model 36703.330 5794.612 2607.433 28301.285

Cost-sharing contract model 37088.093 8903.334 5725.358 22459.401

D
TE

In Table 2, we can find that both the retail prices and wholesale prices of two products
obtain the highest in the cost-sharing contract model, followed by the MS-Stackelberg model
and obtain the lowest in the RS-Bertrand model. It is obvious that the retail prices in the
EP

Stackelberg model are higher than those in the Bertrand model. We also derive an interesting
result: the green product (i.e., product 1) achieves higher retail prices in MS game than those
C

in RS Game. As we know, during the early periods of green product sales, the retailer usually
uses the relatively lower prices to attract customers and expand the market. On the other hand,
AC

there is a small difference between MS game and RS game for product 2’s retail prices. In
addition, the green manufacturing level obtains the highest in the centralized model and
obtains the lowest in the MS-Bertrand model. Moreover, the green manufacturing levels are
higher in RS game than those in MS game.
In Table 3, we find that the profits of the entire supply chain achieve the highest in the
centralized model and reach the lowest in the MS-Stackelberg model. Moreover, the RS game
can make the supply chain obtain more profits than the MS game can. The retailer achieves
the highest profits in the RS-Bertrand model, followed by the RS-Stackelberg model and

16
 ACCEPTED MANUSCRIPT
obtains the lowest profits in the MS-Stackelberg model.
Based on the above results, we can gain some interesting insights. The centralized model
can make the green manufacturing level and the supply chain profits achieve the highest,
while the retail prices and the wholesale prices obtain the highest in the cost-sharing contract
model. Moreover, when the retailer is the Stackelberg leader in the RS game, the retailer
obtains more profits under the Bertrand model than under the Stackelberg model.
7.2. The impact of green investment on the green manufacturing level and supply chain

PT
performance
We now further study the effects of green investment parameter on the optimal green

RI
manufacturing level and supply chain performance under different models. For the numerical
study, we assume that a=800, ε=0.6, c1=16, c2=15, α =6, γ=4, β=3, ϕ=0.6 and the value of I

SC
varies from 2 to 10.
7.2.1. The impact of green investment on the green manufacturing level
Fig. 2 illustrates the effect of the green investment parameter (I) on the green

U
manufacturing level in the six models as indicated in the figure. It shows that the optimal
AN
green manufacturing level decreases with the green investment parameter, and the green
manufacturing levels under the six models tend to be closer when the green investment
parameter is relatively large. Moreover, the green manufacturing level obtains the highest in
M

the centralized model and reached the lowest in the MS-Bertrand model. Fig. 2 also shows
that the green manufacturing level in RS game is better than in MS game, and the green
D

manufacturing level in the Stackelberg model is better than the Bertrand model.
For enterprises, costs and risks are two issues they need to face. There is not necessarily
TE

a directly proportional relationship between inputs and outputs. As a rational being, each
member of the supply chain focuses on maximizing its own profits. Thus, most of them would
EP

rather adopt a wait-and-see attitude or invest less in the early stage of green production. As a
result, the centralized model is more conducive to enhance green manufacturing level than
other five models.
C
AC

17
 ACCEPTED MANUSCRIPT

PT
RI
Fig.2.The effect of green investment parameter (I) on the green manufacturing level

SC
7.2.2. The impact of green investment on supply chain performance
Figs. 3-6 demonstrate that the effect of the green investment parameter (I) on the profits

U
of the supply chain, retailer, M1 and M2, respectively. Fig. 3 shows that the supply chain
profits decrease with the green investment parameter. Additionally, we also derive some
AN
interesting results in Figs. 3-6. First, we find that the supply chain profits are higher in the
cost-sharing contract model than those in the decentralized models. Second, the
M

manufacturers gain more profits under the Stackelberg models than under the Bertrand
models. On the contrary, the retailer obtains more profits under the Bertrand models.
Moreover, the profits of M2 decrease with the green investment parameter in MS models
D

while increasing with the green investment parameter in RS models. Later, both of the
TE

decision scenarios tend to become stable. The explanations for these phenomena can be given
as follows.
Firstly, we find that Figs. 3-5 first have a declining trend and then become steady.
EP

Second, it is favorable for M1 to share the green manufacturing cost on account of

transferring various risks appropriately and then maintain an intimate relationship with the
C

retailer. Moreover, the cost-sharing contract model can improve supply chain performance
than other decentralized models. We also find that supply chain profits under the centralized
AC

case are higher than those in the cost-sharing contract and other decentralized models, while
the cost-sharing contract model will be better for the M1, the retailer as well as the supply
chain than the other decentralized models.

18
 ACCEPTED MANUSCRIPT

PT
RI
Fig.3. The effect of green investment parameter (I) on the supply chain profits

U SC
AN
M
D
TE

Fig.4. The effect of green investment parameter (I) on the retailer’s profits
C EP
AC

Fig.5. The effect of green investment parameter (I) on the M1’s profits

19
 ACCEPTED MANUSCRIPT

PT
RI
Fig.6. The effect of green investment parameter (I) on the M2’s profits

SC
7.2.3. Comparison between the profits of M1 and M2

U
Recall that M1 and M2 manufacture products substitutable to each other. The difference
AN
between M1 and M2 is that M1 invests in green manufacturing. Therefore, product 1 is much

greener and more expensive than product 2. Assume that a=800, ε=0.6, c1=16, c2=15, α =6,
M

γ=4, β =3, and the value of I varies from 2 to 10. Figs. 7-8 and 11 demonstrate that both the

profits of M1 and M2 decrease with green investment parameter (I). Figs. 9-10 show that
D

M1’s profits decrease with the green investment parameter (I) while M2’s profits increase
TE

with the green investment parameter (I).

As the green investment parameter (I) increases, we find that M1’s profits are higher

than those of M2, which indicates that M1 can benefit from green manufacturing. Although
EP

green manufacturing will incur more costs to M1, the demand for its green product will also

increase. Usually, customers are willing to buy a new, different or novel product if its price is
C

acceptable. Because of its green nature, product 1 will attract many consumers to buy and the
AC

demand will increase accordingly. Thus, in its initial stage, green manufacturing provides M1

with great benefit and a competitive advantage over M2. Nonetheless, this does not mean that

the more investment M1 makes in green manufacturing, the more profit it will obtain. When

the green manufacturing level has achieved a steady state or the maximum, it is unnecessary

to invest more in green manufacturing. Suppose that M1 continues green investment, it has to

raise the wholesale price of product 1 to ensure profit. This will cause the retailer to raise the

retail price of product 1, leading to a decrease in demand of the product. This is the reason
20
 ACCEPTED MANUSCRIPT

why M1 is more competitive and profitable than M2 at the beginning of green manufacturing.

PT
RI
SC
Fig.7.The effect of green investment parameter (I) on the manufacturers' profits (MS-Bertrand
model)

U
AN
M
D
TE
EP

Fig.8.The effect of green investment parameter (I) on the manufacturers' profits

(MS-Stackelberg model)
C
AC

21
 ACCEPTED MANUSCRIPT
Fig.9.The effect of green investment parameter (I) on the manufacturers' profits (RS-Bertrand
model)

PT
RI
SC
Fig.10.The effect of green investment parameter (I) on the manufacturers' profits

U
(RS-Stackelberg model)
AN
M
D
TE
EP

Fig.11.The effect of green investment parameter (I) on the manufacturers’ profits in

cost-sharing contract model (߶=0.6)
C
AC

7.3. The impact of parameter α, γ and β on the green manufacturing level

We now study the impact of parameters α, γ and β on the green manufacturing level.
Using the same parameters as in Section 7.1: a=800, ε=0.6, c1=16, c2=15 and let I=2 in this
section.
7.3.1. The impact of parameter α on the green manufacturing level
To study the relationship between price elasticity (α) and the green manufacturing level
(θ), we set γ=4, β=3 and α∈[6,10]. Fig. 12 shows that the green manufacturing level decreases
with price elasticity α. Specially, in the six models, the green manufacturing level achieves the

22
 ACCEPTED MANUSCRIPT
highest in the centralized model and reaches the lowest in the MS-Bertrand model. Moreover,
the effect of price elasticity (α) on the green manufacturing level is not obvious when
parameter α is relatively large.

PT
RI
U SC
Fig.12.The effect of price elasticity (α) on the green manufacturing level
7.3.2. The impact of parameter γ on the green manufacturing level
AN
To study the relationship between cross-price sensitivity (γ) and the green manufacturing
level (θ), we denote α=6, β=3 and γ∈[0,4]. Fig. 13 shows that the green manufacturing level
M

increases with cross-price sensitivity (γ). In the centralized model, the green manufacturing
level increases most rapidly with parameter γ, while the green manufacturing level increases
with γ most slowly under the MS-Bertrand model.
D
TE
C EP
AC

Fig.13. The effect of cross-price sensitivity (γ) on the green manufacturing level

7.3.3. The impact of parameter β on the green manufacturing level

To study the relationship between the green manufacturing coefficient (β) and the green
manufacturing level θ, and we set α=6, γ=4 and β∈[0,4]. Fig. 14 shows that the green
manufacturing level increases with parameter β. Particularly, the green manufacturing level
23
 ACCEPTED MANUSCRIPT
achieves the highest in the centralized model and reaches the lowest in the MS-Bertrand
model. Moreover, the green manufacturing level increases with β more quickly under the
centralized model than other models.

PT
RI
U SC
Fig.14. The effect of the green manufacturing coefficient (β) on the green manufacturing level
7.4. The effect of cost-sharing proportion (߶) on the supply chain performance
AN
By setting a=800, ε=0.6, c1=16, c2=15, α=6, γ=4, β=3, I=2, we compare the profits of the

retailer, M1 and M2 respectively in the MS-Bertrand and cost-sharing contract models when
M

the value of the cost-sharing proportion ϕ varies from 0.2 to 1. Fig.15 shows that the retailer’s

profits are higher under the cost-sharing contract model than those under the MS-Bertrand
D

model when parameter ϕ is in a certain range. When the value of ϕ is relatively small, the
TE

retailer’s profits under the cost-sharing contract model are less than those under the

MS-Bertrand model. In this case, the retailer would not accept M1’s contract. However, when
EP

the value of ϕ is in the range of [0.3,0.4], the retailer’s profits increase with ϕ and are larger

than those under the MS-Bertrand model. In this case the retailer would accept the contract.
C

Figs. 16 and 17 shows that both the profits of M1 and M2 under the cost-sharing contract
AC

model are higher than those under the MS-Bertrand model. Besides, the profits of both M1

and M2 will always decrease with ϕ, which indicates that M1 wishes the retailer to share

more of the green manufacturing cost. From the above analysis, we can see that the

cost-sharing contract model will benefit all the supply chain members.

Recall that the two manufacturers are the leaders and the retailer is the follower in the

MS-Bertrand model. Green manufacturing will incur variable costs and may generate new

potential risks. When the retailer shares M1’s green manufacturing cost, M1 will probably

24
 ACCEPTED MANUSCRIPT

enhance the green manufacturing level, which benefits both the retailer and M1. However,

when the cost-sharing proportion ϕ is relatively too small (i.e., the retailer bears more of the

green manufacturing cost), the retailer obtains less profits under the cost-sharing contract

model than those under the MS-Bertrand model. In this case, M1 should determine

appropriate contract parameters to motivate the retailer, or they can set contract parameters

PT
through bargaining.

RI
Optimal retailer's profits

U SC
AN
M

Fig.15. Comparison of the retailer’s profits between the MS-Bertrand model and the
D

cost-sharing contract model

TE
C EP Optimal M1's profits
AC

Fig.16. Comparison of M1’s profits between the MS-Bertrand model and the cost-sharing

contract model

25
 ACCEPTED MANUSCRIPT

Optimal M2's profits

PT
RI
SC
Fig.17. Comparison of M2’s profits between the MS-Bertrand model and the cost-sharing

contract model

U
8. Conclusions
AN
In this study, we address the issue of pricing strategies for a supply chain with two
manufacturers and one retailer, in which market demand is influenced by the prices of the two
different products and the green manufacturing level.
M

First, we investigate and optimize the centralized model. Then, we consider the four
decentralized models. In addition, based on the MS-Bertrand model, we consider the
D

cost-sharing contract model. We find that the centralized model (Model I) is superior to the
other models, followed by the cost-sharing model (Model VI), while the MS-Bertrand model
TE

(Model II) has the lowest efficiency. Moreover, the manufacturers can obtain more profits
under the Stackelberg model than those under the Bertrand model, while the retailer can
EP

achieve more profits under the Bertrand model than those under the Stackelberg model. To
improve supply chain performance, both the manufacturers and the retailer must have mutual
trust and share the green manufacturing cost.
C

Subsequently, we analyze the effects of the green investment parameter on the green
AC

manufacturing level and supply chain performance. We also compare the profits of M1 and
M2 under Model II, Model III, Model IV, Model V and Model VI respectively, and find that
M1’s profits are higher than M2’s profits. The results indicate that green manufacturing is
encouraging and worth of popularizing. Then we study the impacts of parameter α, γ and β on
the green manufacturing level. We find that the centralized model is the best, followed by the
cost-sharing contract model, and the MS-Bertrand model is the worst.
For future research, we can study the coordination problem with green manufacturing.
Moreover, we can investigate the pricing strategies under information asymmetry of the green
investment parameter of M1.
26
 ACCEPTED MANUSCRIPT
Acknowledgments
We sincerely thank the editor and the anonymous reviewers for their constructive and
important comments on this paper. This work is supported by the National Natural Science
Foundation of China (Grant Nos.71601099 and 71572040), the Natural Science Foundation of
the Jiangsu Province (BK20160973) and the Humanity and Social Science Youth Foundation
of the Ministry of Education of China (15YJC630091). The work of X. Hong was supported

PT
in part by the National Natural Science Foundation of China (Grant No. 71672071) and the
Fundamental Research Funds for the Central Universities (Grant No. 2662017PY044).

RI
References

SC
Ansari, Z. N., Kant, R., 2017. A state-of-art literature review reflecting 15 years of focus on
sustainable supply chain management. J. Clean. Prod. 142: 2524-2543.
Bajwa, N., Sox, C. R., Ishfaq, R., 2016. Coordinating pricing and production decisions for

U
multiple products. Omega-Int. J. Manage. S. 64: 86-101.
AN
Bechtsis, D., Tsolakis, N., Vlachos, D., Iakovou, E., 2017. Sustainable supply chain
management in the digitalisation era: The impact of Automated Guided Vehicles. J.
Clean. Prod. 142: 3970-3984.
M

Bhattacharya, A., Dey, P. K., Ho, W., 2015. Green manufacturing supply chain design and
operations decision support. Int. J. Prod. Res. 53(21): 6339-6343.
D

Bish, E. K., Suwandechochai, R., 2010. Optimal capacity for substitutable products under
operational postponement. Eur. J. Oper. Res. 207: 775-783.
TE

Cao, J., Zhang, X.M., 2013. Coordination strategy of green supply chain under the free
market mechanism. Energy Procedia 36: 1130-1137.
EP

Ceryan, O., Sahin, O., Duenyas I., 2013. Dynamic pricing of substitutable products in the
presence of capacity flexibility. Manuf. Serv. Oper. Manag. 15 (1): 86-101.
Chen, Y. C., Fang, S. C., Wen, U. P., 2013. Pricing policies for substitutable products in a
C

supply chain with Internet and traditional channels. Eur. J. Oper. Res. 224: 542-551.
AC

Chen, Y. J., Sheu, J. B., 2009. Environmental-regulation pricing strategies for green supply
chain management. Transp. Res. Part E: Logist. Transp. Rev. 45(5): 667-677.
Chin, T. A., Tat, H. H., Sulaiman, Z., 2015. Green supply chain management, environmental
collaboration and sustainability performance. Procedia CIRP. 26: 695-699.
Dai, R., Zhang, J., Tang, W. S., 2017. Cartelization or cost-sharing? Comparison of cooperation
modes in a green supply chain. J. Clean. Prod.156: 159-173.
Fahimnia, B., Sarkis, J., Boland, J., Goh, M., 2015. Policy insights from a green supply chain
optimisation model. Int. J. Prod. Res. 53(21): 6522-6533.
Geng, R., Mansouri, S. A., Aktas, E., 2017. The relationship between green supply chain
27
 ACCEPTED MANUSCRIPT
management and performance: A meta-analysis of empirical evidences in Asian
emerging economies. Int. J. Prod. Econ. 183: 245-258.
Ghosh, D., Shah, J., 2015. Supply chain analysis under green sensitive consumer demand and
cost sharing contract. Int. J. Prod. Econ. 164: 319-329.
Gürler, Ü., Yılmaz, A., 2010. Inventory and coordination issues with two substitutable
products. Appl. Math. Model. 34: 539-551.
Hafezalkotob, A., 2017. Competition, cooperation, and coopetition of green supply chains

PT
under regulations on energy saving levels. Transp. Res. Part E: Logist. Transp. Rev. 97:
228-250.

RI
Hafezalkotob, A., 2015. Competition of two green and regular supply chains under
environmental protection and revenue seeking policies of government. Comput. Ind. Eng.

SC
82: 103-114.
Hsieh, C. C., Wu, C. H., 2009. Coordinated decisions for substitutable products in a common
retailer supply chain. Eur. J. Oper. Res. 196: 273-288.

U
Jabbour, A. B. L. S., Jabbour, C. J. C., Latan, H., Teixeira, A. A., Oliveira ,J. H. C., 2015.
AN
Reprint of ‘‘Quality management, environmental management maturity, green supply
chain practices and green performance of Brazilian companies with ISO 14001
certification: Direct and indirect effects’’. Transp. Res. Part E: Logist. Transp. Rev. 74:
M

139-151.
Jia, J.X., Zhang, J., 2013. Dynamic ordering and pricing strategies in a two-tier
D

multi-generation durable goods supply chain. Int. J. Prod. Econ. 144(1): 135-142.
Karakul, M., Chan, L. M. A., 2010. Joint pricing and procurement of substitutable products
TE

with random demands-A technical note. Eur. J. Oper. Res. 201: 324-328.
Kuei, C. H., Madu, C. N., Chow, W. S., Chen, Y., 2015. Determinants and associated
EP

performance improvement of green supply chain management in China. J. Clean. Prod.

95: 163-173.
Kumara, V., Holt, D., Ghobadian, A., Garza-Reyes, J. A., 2015. Developing green supply
C

chain management taxonomy-based decision support system. Int. J. Prod. Res. 53(21):
AC

6372-6389
Kunter, M., 2012. Coordination via cost and revenue sharing in manufacturer–retailer
channels. Eur. J. Oper. Res. 216: 477-486.
Laari, S., Töyli, J., Solakivi, T., Ojala, L., 2016. Firm performance and customer-driven green
supply chain management. J. Clean. Prod. 112: 1960-1970.
Laari, S., Töyli, J., Ojala, L., 2017. Supply chain perspective on competitive strategies and green
supply chain management strategies. J. Clean. Prod. 141: 1303-1315.
Li, B., Zhu, M., Jiang, Y., Li, Z., 2016. Pricing policies of a competitive dual-channel green
supply chain. J. Clean. Prod. 112: 2029-2042.
28
 ACCEPTED MANUSCRIPT
Li, J., Pan, S. Y., Kim, H., Linn, J. H., Chiang, P. C., 2015. Building green supply chains in
eco-industrial parks towards a green economy: Barriers and strategies. J. Environ.
Manag. 162: 158-170.
Liu, P., Yi, S., 2017. Pricing policies of green supply chain considering targeted advertising
and product green degree in the Big Data environment. J. Clean. Prod. 164: 1614-1622.
Ma, C.S., Li, T., Li, C., He, H., 2014. Pricing strategy and governments intervention for green
Supply chain with strategic customer behavior. J. Syst. Sci. Inf. 2(3): 206-216.

PT
Ma, P., Wang, H., Shang, J., 2013. Contract design for two-stage supply chain coordination:
Integrating manufacturer-quality and retailer-marketing efforts. Int. J. Prod. Econ.

RI
146:745-755.
Seman, N. A. A., Zakuan, N., Jusoh, A., Saman, M.Z.M., 2012. The relationship of green

SC
supply chain management and green innovation concept. Procedia-Soc. Behav. Sci.
57:453-457.
Şen, A., 2016. Competitive markdown timing for perishable and substitutable products.

U
Omega 64: 24-41.
AN
Sharma, V. K., Chandna, P., Bhardwaj, A., 2017. Green supply chain management related
performance indicators in agro industry: A review. J. Clean. Prod. 141: 1194-1208.
Shin, H., Park, S., Lee, E., Benton, W. C., 2015. A classification of the literature on the
M

planning of substitutable products. Eur. J. Oper. Res. 246: 686-699.

Sulistio, J., Rini, T. A., 2015. A structural literature review on models and methods analysis of
D

green supply chain management. Procedia Manuf. 4: 291-299.

Tang, C.S., Yin, R., 2007. Joint ordering and pricing strategies for managing substitutable
TE

products. Prod. Oper. Manag. 16(1):138-153.

Tsao, Y. C., Sheen, G. J., 2012. Effects of promotion cost sharing policy with the sales
EP

learning curve on supply chain coordination. Comp. Oper. Res. 39: 1872-1878.
Ülkü, M. A., Hsuan, J., 2017. Towards sustainable consumption and production Competitive
pricing of modular products for green consumers. J. Clean. Prod. 142: 4230-4242.
C

Wang, Q., Zhao, D., He, L., 2016. Contracting emission reduction for supply chains
AC

considering market low-carbon preference. J. Clean. Prod. 120: 72-84.

Zhang, L., Wang, J., You, J., 2015. Consumer environmental awareness and channel
coordination with two substitutable products. Eur. J. Oper. Res. 241: 63-73.
Zhang, Q., Tang, W., Zhang, J., 2016. Green supply chain performance with cost learning and
operational inefficiency effects. J. Clean. Prod. 112: 3267-3284.
Zhao, J., Wei, J., Li, Y., 2014. Pricing decisions for substitutable products in a two-echelon
supply chain with firms' different channel powers. Int. J. Prod. Econ. 153: 243-252.
Zhao, J., Tang, W., Wei, J., 2012. Pricing decision for substitutable products with retail
competition in a fuzzy environment. Int. J. Prod. Res. 135: 144-153.
29
 ACCEPTED MANUSCRIPT

PT
Appendix

RI
Proof of Proposition 1. In the centralized case, we solve the objective function of the supply
chain.

SC
The first order derivative of Eq. (3) with respect to p 1 , p 2 and θ can be set to 0 as
follows:

U
∂ΠSC
= ε a − α p1 + γ p2 + βθ − ( p1 − c1 )α + ( p2 − c2 )γ =0 , (A1)
∂p1
AN
∂ΠSC
= (1 − ε )a + γ p1 − α p2 + ( p1 − c1 )γ − ( p2 − c2 )α =0 , (A2)
∂p2
M

∂ΠSC
= ( p1 − c1 ) β − 2 Iθ =0 . (A3)
∂θ
D

We can also obtain the Hessian matrix of Π SC as follows:

TE

 ∂ 2 Π SC ∂ 2 Π SC ∂ 2 Π SC 
 
 ∂p1
2
∂p1 p2 ∂p1θ 
 ∂2Π  −2α 2γ β 
∂ 2 Π SC ∂ 2 Π SC   
H（p1 , p2 ,θ ) =   = 2γ −2α
EP

SC
0 . (A4)
 ∂p2 p1 ∂p22 ∂p2θ  
 2   β 0 −2 I 
 ∂ Π SC ∂ 2 Π SC ∂ 2 Π SC 
 ∂θ p ∂θ p2 ∂θ 2 

C

Note that the Hessian matrix of Π SC is a negative definite for p1 , p2 and θ if I and
AC

β satisfy the conditions 4 I (α 2 − γ 2 ) − αβ 2 > 0 .

Combining Eqs. (A1)-(A3), we can obtain:
2 IA2 − c1αβ 2
p1I * = , (A5)
4 IA1 − αβ 2

4 IA3 − A4 β 2
p2I * = , (A6)
2(4 IA1 − αβ 2 )

30
 ACCEPTED MANUSCRIPT
A5 β
θ I* = , (A7)
4IA1 − αβ 2
Substituting Eqs. (A5)-(A7) into Eq. (3), we can obtain the optimal profit of the supply
chain:
4 IA6 − ( A7 β )2
Π SC
I*
= . (A8)
4(4 IA1 − αβ 2 )

PT
where

A1 = α 2 − γ 2 , A2 = (α − γ )(aε + c1α + c1γ ) + aγ , A3 =(α − γ )(− aε + c2α + c2γ ) + aα ,

RI
A4 = a − aε + c2α + c1γ , A5 = (α − γ )(aε − c1α − c1γ ) + aγ ,

SC
A6 = (α 2 − γ 2 )(−2ac1ε + 2ac2ε − 2c1c2γ − 2ac2 + α c12 + α c22 ) − 2ε (α − γ )(1 − ε )a 2 + α a 2 ,

A7 = a − aε − c2α + c1γ .

U
Proof of Proposition 2. Taking the first derivative of Eq. (6) with respect to β, we obtain
AN
∂θ I * (c1γ 2 − c1α 2 + aεα − aεγ + aγ )(4 Iα 2 − 4 I γ 2 + αβ 2 )
= .
∂β (4 Iα 2 − 4 I γ 2 − αβ 2 ) 2

Recall that α>γ>0, 4 Iα 2 − 4 I γ 2 − αβ 2 > 0 , and

M

∂θ I *
c1γ 2 − c1α 2 + aεα − aεγ + aγ = (α − γ )(aε − c1α − c1γ ) + aγ =A5 , then we have ≥ 0.
∂β
D

Proof of Proposition 3: In the decentralized case, we solve the objective function of two
TE

manufacturers and the retailer respectively. We first solve the MS-Bertrand model.
Taking the first order derivative of Eq. (10) with respect to p1 and p2 and setting
EP

them equal to 0, we obtain that:

∂ΠR
= aε − α p1 + γ p2 − ( p1 − ω1 )α + ( p2 − ω2 )γ + βθ = 0 , (A9)
∂p1
C

∂ΠR
= (1 − ε )a − α p2 + γ p1 + ( p1 − ω1 )γ − ( p2 − ω2 )α = 0 , (A10)
AC

∂p2
and the Hessian matrix of Π R is as follows:

 ∂ 2Π R ∂ 2Π R 
 
∂p 2 ∂p1 p2   −2α 2γ 
H ( p1 , p2 ) =  2 1 = . (A11)
∂ Π ∂ 2 Π R   2γ −2α 
 R

∂p
 2 1 p ∂p22 

Eq. (A11) shows that the Hessian matrix of Π R is a negative definite for all values of

p1 and p2 if α > γ . From Eqs. (A9) and (A10), we can derive that:

31
 ACCEPTED MANUSCRIPT
aαε − aγε + ω1α 2 − ω1γ 2 + αβθ + aγ
p1 = , (A12)
2(α 2 − γ 2 )

−aαε + aγε + ω2α 2 − ω2γ 2 + γβθ + aα

p2 = . (A13)
2(α 2 − γ 2 )
Substituting Eqs. (A12) and (A13) into Eqs. (8) and (9), we can obtain that:

(γω1ω2 − c1γω2 − aε c1 + aεω1 + αc1ω1 − αω12 − βθ c1 + βθω1 )

ΠM1 = − Iθ 2 , (A14)
2

PT
(ω 2 − c2 )( a − aε + γω1 − αω 2 )
Π M2 = . (A15)
2

RI
Taking the first order derivative of Eq. (A14) with respect to ω1 and θ , and the first

order derivative of (A15) with respect to ω2 and setting them equal to 0, we can obtain that:

SC
∂Π M1 γω2 βθ aε + α c1
= −αω1 + + + =0, (A16)
∂ω1 2 2 2

U
∂Π M 2 γω1 a − aε + α c2
= − αω2 + =0, (A17)
∂ω2 2 2
AN
∂ΠM1 (ω1 − c1 )β
= − 2Iθ =0 . (A18)
∂θ 2
M

Accordingly, the Hessian matrix of ΠM1 is as follows:

 ∂ 2 Π M1 ∂ 2 Π M1   β 
D

  −α
 ∂ω1
2
∂ω1θ   2 
H (ω1 ,θ ) =  2 = . (A19)
∂ 2 Π M1   β
TE

 ∂ Π M1  −2 I 
 ∂θω ∂θ 2   2 
 1

The Hessian matrix of ΠM1 is a negative definite for all ω1 and θ if I, α and β
EP

β2
satisfy the conditions that 2Iα − > 0.
4
C

Then the second-order derivatives of ΠM2 are given below:

AC

∂ 2Π M 2
= −α . (A20)
∂ω 22

∂ΠM2
Because Eq. (A20) is negative, thus =0 has the unique solution.
∂ω2
Solving Eqs. (A16)-(A18), we find that
2 IB2 − c1αβ 2
ω1II * = , (A21)
IB1 − αβ 2

32
 ACCEPTED MANUSCRIPT
4 IB3 − A4 β 2
ω 2II * = , (A22)
2( IB1 − αβ 2 )

B4 β
θ II * = . (A23)
2( IB1 − αβ 2 )
Substituting Eqs. (A21)-(A23) into Eqs. (A12), (A13), (8)-(10), we obtain that
4 IB5 − B6 β 2
p1II * = , (A24)
4 A1 ( IB1 − αβ 2 )

PT
4 IB7 − B8 β 2
p2II * = , (A25)
4 A1 ( IB1 − αβ 2 )

RI
I (8Iα − β 2 ) B42
Π MII *1 = , (A26)
4( IB1 − αβ 2 )2

SC
α (4 IB9 − A7 β 2 )2
Π MII *2 = , (A27)
8( IB1 − αβ 2 )2

[4 IB4 (4 IB10 − A7γβ 2 ) + (4 IB9 − A7 β 2 )(4 IB11 − A7αβ 2 )]α 2

U
Π IIR * = . (A28)
16 A1 ( IB1 − αβ 2 ) 2
AN
Thus, the total profit of the supply chain is as follows:
Π SC
II *
= Π MII *1 + Π MII *2 + Π IIR *
M

[4 IB4 (4 IB10 − A7γβ 2 ) + (4 IB9 − A7 β 2 )(4 IB11 − A7αβ 2 )]α 2

+ A1[4 I (8Iα − β 2 ) B42 + 2α (4 IB9 − A7 β 2 ) 2 ]
= . (A29)
D

16 A1 ( IB1 − αβ 2 )2
where
TE

B1 = 8α 2 − 2γ 2 , B2 = 2aαε − aγε + 2c1α 2 + c2αγ + aγ ,

EP

B3 = −2aαε + aγε + 2c2α 2 + c1αγ + 2aα , B4 = aε (2α − γ ) − c1 (2α 2 − γ 2 ) + γ (a + c2α) ,

B5 = aγ (1− ε )(5α 2 − 2γ 2 ) + α(2c1α + c2γ )(α 2 − γ 2 ) + 3aαε (2α 2 − γ 2 ) ,

C

B6 = αγ (a − aε − c2α) + c1α(4α 2 − 3γ 2 ) ,
AC

B7 = aγε (5α 2 − 2γ 2 ) + α(2c2α + c1γ )(α 2 − γ 2 ) + 3aα(1− ε )(2α 2 − γ 2 ) ,

B8 = (a − aε + c1γ )(3α 2 − 2γ 2 ) + c2α(α 2 − 2γ 2 ) ,

B9 = −aε (2α − γ ) − c2 (2α 2 − γ 2 ) + α(2a + c1γ ) ,

B10 = aε (2α − γ )(α − γ ) − (α 2 − γ 2 )(2c1α + c2γ ) + 3aαγ ,

B11 = −aε (2α − γ )(α − γ ) − (α 2 − γ 2 )(2c2α + c1γ ) + a(2α 2 + γ 2 ) .

Proof of Proposition 4. Taking the first derivative of Eq. (13) with respect to β, we get

33
 ACCEPTED MANUSCRIPT
∂θ II * (c1γ 2 − 2c1α 2 + 2aαε − aεγ + c2αγ + aγ )(8 Iα 2 − 2 I γ 2 + αβ 2 )
= .
∂β 2(8Iα 2 − 2 I γ 2 − αβ 2 )2

β2
Recall that α >γ , 2Iα − >0 , 8Iα 2 − 2 I γ 2 + αβ 2 > 0 , and
4
c1γ 2 − 2c1α 2 + 2aαε − aεγ + c2αγ + aγ =aε (2α − γ ) − c1 (2α 2 − γ 2 ) + γ (a + c2α)=B4 > 0 . Thus,
∂θ II *
we have ≥ 0.
∂β

PT
Proof of Proposition 5: In the MS-Stackelberg model, two manufacturers make sequence
decisions. Similar to the MS-Bertrand model, we first derive the response functions of the

RI
retailer, i.e., Eqs. (A12) and (A13).
Then, we solve Eqs. (A17) and get:

SC
γω1 + a − aε + α c2
ω2 = . (A30)
2α
Substituting Eqs. (A12), (A13) and (A30) into Eq. (8), we can obtain that:

Π M1 = −
1
U
[(2α 2 − γ 2 )ω12 + 4 I αθ 2 + 2c1αβθ − aεγ c1 + 2 aαε c1 + αγ c1c2 + aγ c1
AN
4α . (A31)
+ ( aεγ − 2 aαε − 2α c1 + γ c1 − αγ c2 − aγ − 2αβθ )ω1 ]
2 2

Taking the first order derivative of Eq. (A31) with respect to ω 1 and θ, and setting
M

them equal to 0, we can obtain that:

∂Π M1 (−4α 2 + 2γ 2 )ω1 + 2αβθ + (2α 2 − γ 2 )c1 + αγ c2 + aγ (1 − ε ) + 2aαε
D

= = 0, (A32)
∂ω1 4α
TE

∂ΠM1 βω1 β c1
= − 2Iθ − = 0. (A33)
∂θ 2 2
The Hessian matrix of ΠM1 is as follows:
EP

 ∂ 2 Π M1 ∂ 2 Π M1   −2α 2 + γ 2
  β 
 ∂ω1
2
∂ω1θ   
C

H (ω1 ,θ ) =  2 = 2α 2 
 . (A34)
β
 ∂ Π M1 ∂ Π M1   
2

 −2 I 
AC

 ∂θω 2 
∂θ   2 
 1

The Hessian matrix of ΠM1 is a negative definite for all ω1 and θ if I , α and β

8Iα 2 − 4γ 2 − αβ 2
satisfy the conditions that >0.
4α
Solving Eqs. (A32)-(A33), we find that
2 IB13 − c1αβ 2
ω1III * = , (A35)
IB12 − αβ 2

34
 ACCEPTED MANUSCRIPT
B4 β
θ III * = . (A36)
2( IB12 − αβ 2 )

After substituting Eq. (A35) into Eq. (A30), we can find that
2 IB14 − A4αβ 2
ω2III * = . (A37)
2α ( IB12 − αβ 2 )
Substituting Eqs. (A35)-(A37) into Eqs. (A12), (A13), (8)- (10), we attain that:

PT
4 IB15 − B6 β 2
p1III * = , (A38)
4 A1 ( IB12 − αβ 2 )

2 IB16 − B17 β 2

RI
p2III * = , (A39)
2α A1 ( IB12 − αβ 2 )

IB42

SC
Π III
M1 =
*
, (A40)
4α ( IB12 − αβ 2 )

(2 IB18 − A7αβ 2 ) 2
Π MIII2* = , (A41)
8α ( IB12 − αβ 2 ) 2

U
4 B19 I 2 + 4 B20αβ 2 I + A72α 4 β 4
AN
Π III *
= . (A42)
16α A1 ( IB12 − αβ 2 ) 2
R

Thus, the total profit of the supply chain can be expressed as follows:
M

ΠSC
III *
= ΠMIII1* + ΠMIII2* + ΠRIII * . (A43)

where
D

B12 = 8α2 − 4γ 2 , B13 = aε (2α − γ ) + c1(2α 2 − γ 2 ) + γ (a + αc2 ) ,

TE

B14 = (4α 2 − γ 2 )(a − aε + αc2 ) + c1γ (2α 2 − γ 2 ) + 2aαγε ,

B15 = aγ (1 − ε )(5α 2 − 3γ 2 ) + [c1 (2α 2 − γ 2 ) + αγ c2 ](α 2 − γ 2 ) + 2aαε (3α 2 − 2γ 2 ) ,
EP

B16 = (α 2 − γ 2 )[α c2 (4α 2 − γ 2 ) + γ c1 (2α 2 − γ 2 )] + 2 aαγε (5α 2 − 3γ 2 )

,
+ a (1 − ε )(12α 4 − 9α 2γ 2 + γ 4 )
C

B17 = α(a − aε + γ c1)(3α 2 − 2γ 2 ) + c2α 2 (α 2 − 2γ 2 ) ,

B18 = γ c1 (2α 2 − γ 2 ) − α c2 (4α 2 − 3γ 2 ) + a(1 − ε )(4α 2 − γ 2 ) + 2aαγε ,
AC

B19 = +(α 2 − γ 2 )[c22α 2γ 4 + a 2ε 2γ 2 (16α 2 − 3γ 2 ) − 4ac2εγα 2 (8α 2 − 3γ 2 )

−2α c1c2γ (2α 2 − γ 2 )] − 2ac1 (α 2 − γ 2 )(2α 2 − γ 2 )[γ (1 − ε )(8α 2 − 3γ 2 )
+2εα (4α 2 − γ 2 )] + 16aα 6 (2ε 2 − 2ε + 1) + 4a 2αγε (1 − ε )[(4α 2 − γ 2 )2 ,
+α γ ] + c (α − γ )(4α − 3γ )(2α − γ ) − 2ac2α (1 − ε )(16α + γ
2 2 2
1
2 2 2 2 2 2 2 6 6

−20α 4γ 2 + 3α 2γ 4 ) + (α 2 − γ 2 )(2α 2 − γ 2 )[3a 2γ 2 (1 − 2ε ) + 4α 2c22 (α 2 − γ 2 )]

35
 ACCEPTED MANUSCRIPT
B20 = −a 2 (1 − ε )2 (4α 4 + α 2γ 2 − γ 4 ) − 2ac1γ (1 − ε )(α 4 + 2α 2γ 2 − γ 4 )
+4ac2α 3 (1 − ε )(2α 2 − γ 2 ) − 2aεαγ (3α 2 − γ 2 )(a − ε a − c2α + c1γ ) .
−(α − γ )[c α (4α − γ ) − c γ (2α − γ ) − 2c1c2γα ]
2 2 2
2
2 2 2 2
1
2 2 2 3

Proof of Proposition 6. Taking the first derivative of Eq. (22) with respect to β, we get
∂θ III * (c1γ 2 − 2c1α 2 + 2aαε − aεγ + c2αγ + aγ )(8Iα 2 − 4I γ 2 + αβ 2 )
= .
∂β 2(8Iα 2 − 4I γ 2 − αβ 2 )2

8Iα 2 − 4γ 2 − αβ 2

PT
Recall that α >γ , >0 , 8Iα 2 − 4 I γ 2 + αβ 2 > 0 , and
4α
c1γ 2 − 2c1α 2 + 2aαε − aεγ + c2αγ + aγ =aε (2α − γ ) − c1 (2α 2 − γ 2 ) + γ (a + c2α)=B4 > 0 . Thus,

RI
∂θ III *
we have ≥ 0.
∂β

SC
Proof of Proposition 7. In the RS-Bertrand model, we introduce the concept of marginal
profit as the decision variable to substitute the retail price:

U
p1 = ω1 + m1 , (A44)
AN
p2 = ω2 + m2 . (A45)

By substituting Eqs. (A44)-(A45) into Eqs. (8)-(10), we attain that:

ΠM1 = (ω1 − c 1)[ε a − α (ω1 + m1 ) + γ (ω2 + m2 ) + βθ ] − Iθ 2 ,

M

(A46)

ΠM2 = (ω2 − c2 )[(1 − ε ) a − α (ω2 + m2 ) + γ (ω1 + m1 )] , (A47)

D

Π R = m1[ε a − a (ω1 + m1 ) + γ (ω2 + m2 ) + βθ ] + m2 [(1 − ε ) a

. (A48)
− a (ω2 + m2 ) + γ (ω1 + m1 )]
TE

Taking the first order derivative of Eq. (A46) with respect to ω1 and θ , and the first

order derivative of (A47) with respect to ω1 and setting them equal to 0, we can obtain that:
EP

∂Π M1
= −2αω1 + γω2 + βθ − α m1 + γ m2 + aε +α c1 = 0 , (A49)
∂ω1
C

∂Π M 2
= −2αω2 + γω1 − α m2 + γ m1 + a − aε +α c2 = 0 , (A50)
AC

∂ω2

∂ΠM1
= β (ω1 − c1 ) − 2Iθ = 0 . (A51)
∂θ
Therefore, the Hessian matrix of ΠM1 is as follows:

 ∂ 2Π M1 ∂ 2Π M1 
 
 ∂ω1 ∂ω1θ   −2α 2 β 
2

H (ω1 ,θ ) =  2 = . (A52)
 ∂ Π M1 ∂ 2Π M1   β −2 I 
 ∂θω ∂θ 2 
 1

36
 ACCEPTED MANUSCRIPT
The Hessian matrix of ΠM1 is a negative definite for all ω 1 and θ if I , α and β

satisfy the conditions that 4 Iα 2 − β 2 > 0 .

Then we get the second-order derivatives of ΠM2 :

∂ 2Π M 2
= − 2α . (A53)
∂ ω 22

∂ΠM2

PT
Because Eq. (A53) is negative, thus =0 has the unique solution.
∂ω2
Solving Eqs. (A49)-(A51), we obtain that

RI
[m2αγ − (2α 2 − γ 2 )m1 + ε a(2a − γ ) + aγ + α (c2γ + 2c1α )]I − c1αβ 2
ω1 = , (A54)
4 Iα 2 − I γ 2 − αβ 2

SC
1
ω2 = {2[m1αγ − (2α 2 − γ 2 )m2 + 2aα (1 − ε )
2(4Iα − I γ − αβ )
2 2 2
, (A55)
+ε aγ + α (c1γ + 2c2α )]I − [a(1 − ε ) + c2α + c1γ − α m2 + γ m1 )β } 2

U
β [ε a(2α − γ ) − (c1 + m1 )(2α 2 − λ 2 ) + (c2 + m2 )αγ + aγ ]
AN
θ= . (A56)
2(4Iα 2 − I γ 2 − αβ 2 )
Substituting Eqs. (A54)-(A56) into Eq. (A48), we attain that:
M

1
ΠR = {2α I [ε a (2α − γ )( m1 − m2 ) − (2α 2 − γ 2 )(c1m1
2(4 I α − I γ 2 − αβ 2 )
2

+ c2 m2 + m12 + m22 ) + αγ (c1m2 + c2 m1 ) + 2αγ m1m2 + 2aα m2 + aγ m1 ] + αβ 2 [(ε a . (A57)

D

− a + c1γ − c2α ) m2 + α m22 − γ m1m2 ]}

TE

Taking the first order derivative of Eq. (A57) with respect to m 1 and m 2 , and setting

them equal to 0, we can obtain that:

∂Π R
EP

1
= {−2α I [(2m1 + c1 )(2α 2 − γ 2 )
∂m1 2(4Iα − I γ 2 − αβ 2 )
2
, (A58)
−ε a(2α − γ ) − γ (a + α c2 + 2α m2 )] − m2αγβ } = 0
2
C

∂ΠR 1
= {−2α I [(2m2 + c2 )(2α 2 − γ 2 ) − 2aα (1 − ε )
AC

∂m2 2(4Iα − I γ 2 − αβ 2 )
2
. (A59)
−γ (ε a + c1α + 2m1α )] − (aε − a + c2α + 2m2α − c1γ − m1γ )}αβ = 0 2

The Hessian matrix of Π R we can get as follows:

 ∂ 2Π R ∂ 2Π R 
 
∂m 2 ∂m1m2 
H ( m1 , m2 ) =  2 1
∂ Π ∂ 2Π R 
 R

∂ m
 2 1 m ∂m22 

37
 ACCEPTED MANUSCRIPT
 − 2α I (2α 2 − γ 2 ) α γ (4 I α − β 2 ) 
 
4 I α 2 − I γ 2 − αβ 2 2(4 I α − I γ − αβ ) 
2 2 2

= . (A60)
 αγ (4 I α − β 2 ) −α (4 I α 2 − 2 I γ 2 − αβ 2 ) 
 
 2(4 I α − I γ − αβ ) 4 I α 2 − I γ 2 − αβ 2
2 2 2


Eq. (A60) shows that the Hessian matrix of Π R is a negative definite for all values of

α 2 [16 I 2 (α 2 − γ 2 )(4α 2 − γ 2 ) − 16αβ 2 (α 2 − γ 2 ) − γ 2 β 4 ]

m1 and m2 if > 0.
4(4Iα 2 − I γ 2 − αβ 2 )2

PT
Combining Eqs. (A58) and (A59), we can derive that:
8C3 I 2 + C4 I + A7γβ 4
m1IV * =

RI
, (A61)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

2 I (4C5 I − C6 β 2 )
m2IV * = . (A62)

SC
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
Then substituting Eqs. (A61) and (A62) into Eqs. (A54)-(A56), we obtain that:
8C7 I 2 + 2C8 β 2 I − c1γ 2 β 4

U
ω1IV * = , (A63)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
AN
8C9 I 2 + 2C10 β 2 I − c2γ 2 β 4
ω2IV * = , (A64)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
M

(4C11 I + C12γβ 2 ) β
θ IV * = . (A65)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
D

By substituting Eqs. (A63)-(A65) into Eqs. (A44) and (A45), we obtain that:
8C13 I 2 + 2C14 β 2 I + C15
TE

p1IV * = , (A66)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4

8C16 I 2 + 2C17 β 2 I − c2γ 2 β 4

p2IV * = . (A67)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
EP

After substituting Eqs. (A63)-(A67) into Eqs. (8)-(10), we get that:

I (4C11 I + C12γβ 2 ) 2 (4α I − β 2 )
C

Π MIV1* = , (A68)
[16(C1 I 2 + C2 β 2 I ) − γ 2 β 4 ]2
AC

4α I 2 (4C18 I + C19 β 2 )2
Π MIV2* = , (A69)
[16(C1 I 2 + C2 β 2 I ) − γ 2 β 4 ]2

2α I (2C20 I + C21β 2 )
Π IV *
= . (A70)
16(C1 I 2 + C2 β 2 I ) − γ 2 β 4
R

So the total profit of the supply chain is as follows:

ΠSC
IV *
= ΠMIV1* + ΠMIV2* + ΠRIV * . (A71)

where

38
 ACCEPTED MANUSCRIPT
C1 = (α 2 − γ 2 )(4α 2 − γ 2 ) , C2 = −α(α 2 − γ 2 ) , C3 = (4α 2 − γ 2 )[a(εα − εγ + γ ) − c1 (α 2 − γ 2 )] ,

C4 = 2[c1α(4α 2 − 5γ 2 ) + c2γ (2α 2 − γ 2 ) − ε a(4α 2 + γ 2 ) − 6aαγ (1− ε )]β 2 ,

C5 = (4α 2 − γ 2 )[aα(1− ε ) + ε aγ − c2 (α 2 − γ 2 )] ,

C6 = a(1 − ε )(4α 2 − γ 2 ) + [c1γ (2α 2 − γ 2 ) + 2aεαγ ] − c2α (4α 2 − 3γ 2 ) ,

C7 = (α 2 − γ 2 )[aγ (1− ε ) + c1(6α 2 − γ 2 ) + c2αγ + 2aεα] ,

PT
C8 = aγ [εγ + α(1− ε )] − (α 2 − γ 2 )(c2γ + 8c1α) ,

C9 = (α 2 − γ 2 )[2aα(1− ε ) + c2 (6α 2 − γ 2 ) + c1αγ + ε aγ ] ,

RI
C10 = −aεαγ − a(1 − ε )(2α 2 − γ 2 ) − (α 2 − γ 2 )(c1γ + 6c2α ) ,

C11 = (α 2 − γ 2 )[c2αγ − c1 (2α 2 − γ 2 ) + 2aεα + aγ (1 − ε )] , C12 = aα(1− ε ) + aεγ − c2 (α 2 − γ 2 ) ,

SC
C13 = aγ (1− ε )(5α 2 − 2γ 2 ) + α(2c1α + c2γ )(α 2 − γ 2 ) + 3aεα(2α 2 − γ 2 ) ,

U
C14 = −4aεα 2 − 5aαγ (1− ε ) + α(c2αγ − 4c1α 2 + 3c1γ 2 ) , C15 = (a − ε a − c2α )γβ 4 ,
AN
C16 = aεγ (5α 2 − 2γ 2 ) + α(2αc2 + γ c1)(α 2 − γ 2 ) + 3aα(1− ε )(2α 2 − γ 2 ) ,

C17 = −a[2(1− ε )(3α 2 − γ 2 ) + 3εαγ ] − c2α(2α 2 − 3γ 2 ) − c1γ (3α 2 − 2γ 2 ) ,

M

C18 = (α 2 − γ 2 )[c2 (2α 2 − γ 2 ) − 2aα(1 − ε ) − c1αγ − ε aγ ] ,

C19 = a(1 − ε )(2α 2 − γ 2 ) + (c1γ − 2c2α )(α 2 − γ 2 ) + ε aαγ ,

D

C 20 = (α 2 − γ 2 )[(2α 2 − γ 2 )(c12 + c22 ) − 2c1c2αγ ] − 2 a (α 2 − γ 2 )[ε (2α − γ )( c1 − c2 )

,
TE

+ (c1γ + 2c2α )] − 2 a 2ε (1 − ε )(2α − γ )(α − γ ) + a 2 (2α 2 + γ 2 )

C21 = [a(c2α − c1γ ) − a2 (1− ε )][α(1− ε ) + εγ ] + (α 2 − γ 2 )[c1c2γ + ac2 (1− ε ) −αc22 ] .

EP

Proof of Proposition 8. The proof is similar to that of Proposition 2, and is thus omitted here.
Proof of Proposition 9: In the RS-Stackelberg model, similar to the RS-Bertrand model, we
solve Eqs. (A50) and derive the response function of M2 as follows:
C

c2 γω1 + γ m1 − α m2 + a(1 − ε )
ω2 = + . (A72)
AC

2 2α
Substituting Eq. (A72) into Eq. (8), we can obtain that:
1
Π M1 = [(2α 2 − γ 2 )(c1ω1 − m1ω1 − ω12 ) + (2ε aα − ε aγ + c2αγ
2α
+m2αγ + aγ )ω1 + 2αβθ (ω1 − c1 ) − 2α Iθ 2 − c1ε a(2α − γ ) . (A73)
+c1m1 (2α 2 − γ 2 ) − γ (α c1c2 + α c1m2 + ac1 )]

Taking the first order derivative of Eq. (A73) with respect ω 1 and θ, and setting them
equal to 0, we can attain that:

39
 ACCEPTED MANUSCRIPT
∂Π M1 2ε aα − ε aγ + aγ + (2α 2 − γ 2 )[(c1 − m1 ) − 2ω1 ] + 2αβθ + αγ (c2 + m2 )
= = 0 , (A74)
∂ω1 2α

∂ΠM1
= β (ω1 − c1 ) − 2Iθ = 0 . (A75)
∂θ
The Hessian matrix of ΠM1 can be given as follows:

 ∂ 2 Π M1 ∂ 2 Π M1 

PT
   −2α 2 + γ 2 
 ∂ω1 ∂ω1θ   β 
2

H (ω1 ,θ ) =  2 = α
∂ 2Π M1   . (A76)
 ∂ Π M1  β −2 I 
 ∂θω ∂θ 2 

RI
 1

The Hessian matrix of ΠM1 is a negative definite for all ω 1 and θ if I , α and β

SC
4Iα 2 − 2I γ 2 − αβ 2
satisfy the conditions that >0.
α

U
Solving Eqs. (A75) and (A76), we obtain that:
I [2ε aα − ε aγ + aγ + c2αγ + m2αγ + (c1 − m1 )(2α 2 − γ 2 )] − c1αβ 2
ω1 =
AN
, (A77)
4 Iα 2 − 2I γ 2 − αβ 2

β [2ε aα − ε aγ + aγ + c2αγ + m2αγ − (c1 + m1 )(2α 2 − γ 2 )]

θ= . (A78)
2(4Iα 2 − 2I γ 2 − αβ 2 )
M

Substituting Eqs. (A77) and (A78) into Eq.(A72), we attain that:

D

1
ω2 = {I [(4α 2 − γ 2 )( a − aε + c2α ) + 2ε aαγ
2α (4 I α − 2 I γ − αβ )
2 2 2
TE

−α m2 (4α 2 − 3γ 2 ) + (c1 + m1 )(2α 2 − γ 2 )] + (ε aα − aα + m2α 2 − c2α 2 . (A79)

−c1αγ − m1αγ ) β } 2

After substituting Eqs. (A77)-(A79) into Eq.(10), we attain that:

EP

1
ΠR = {I [αγ (c1m2 + c2 m1 + 2m1m2 )(2α 2 − γ 2 )
2α (4 Iα − 2 I γ − αβ )
2 2 2

−α 2 (c2 m2 + m22 )(4α 2 − 3γ 2 ) − (c1m1 + m12 )(2α 2 − γ 2 ) 2 + 2m1ε aα (2α 2 − γ 2 ) . (A80)

C

+(1 − ε )(4m2 aα 3 + 2m1aγα 2 − m2 aαγ 2 − m1aγ 3 )+2m2ε aγα 2 ] + (ε a − a + c2α

AC

−c1γ + m2α − m1γ )m2α 2 β 2 }

Taking the first order derivative of Eq. (A80) with respect to m1 and m2 , and setting

them equal to 0, we can get that:

∂Π R 1
= {I [(2ε aα + 2m2αγ + c2αγ + aγ
∂m1 2α (4Iα − 2I γ 2 − αβ 2 )
2
, (A81)
−ε aγ )(2α − γ ) − (c1 + 2m1 )(2α − γ ) ] − m2γα β } = 0
2 2 2 2 2 2 2

40
 ACCEPTED MANUSCRIPT
∂Π R 1
= {I [γ (c1 + 2m1 )(2α 2 − γ 2 )
∂m2 2(4 I α − 2 I γ 2 − αβ 2 )
2

−α (c2 + 2m2 )(4α 2 − 3γ 2 ) + a (4α 2 − γ 2 )(1 − ε ) + 2ε aαγ ] . (A82)

−[α (c2 + 2m2 ) − aα (1 − ε ) − αγ (c1 + 2m1 )]β } = 0
2 2

The Hessian matrix of Π R is as follows:

 ∂ 2Π R ∂2Π R   − I (2α 2 − γ 2 ) 2 γ 
   

PT
∂m 2 ∂m1m2   α (4 I α 2 − 2 I γ 2 − αβ 2 ) 2
H ( m1 , m2 ) =  2 1 =  . (A83)
∂ Π ∂2Π R   γ −α (4 I α 2 − 3 I γ 2 − αβ 2 ) 
 R
2 
  
∂m
 2 1m ∂m2   2 4 I α 2 − 2 I γ 2 − αβ 2 

RI
Eq. (A83) shows that the Hessian matrix of Π R is a negative definite for all values of

8(α 2 − γ 2 )(2α 2 − γ 2 )[2I 2 (2α 2 − γ 2 ) − Iαβ 2 ] − α 2γ 2 β 4

SC
m1 and m2 if >0.
4(4Iα 2 − 2I γ 2 − αβ 2 )2
Combining Eqs. (A81) and (A82), we can derive that:

U
8C24 I 2 + C25 I + A7γα 2 β 4
m1V * = , (A84)
8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4
AN
I (8C24 I − C26 β 2 )
m2V * = . (A85)
8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4
M

After substituting Eqs. (A84) and (A85) into Eqs. (A77)- (A79), we attain that:
4C27 I 2 + 2C28αβ 2 I − c1α 2γ 2 β 4
ω1V * = , (A86)
D

8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4

2C29 I 2 + C30 β 2 I − c2γ 2α 3 β 4

TE

ω2V * = , (A87)
8α (2C22 I 2 + C23 β 2 I ) − γ 2α 3 β 4

(2C31 I + C8γα 2 β 2 ) β
θ V* = . (A88)
EP

8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4

Substituting Eqs. (A84)- (A88) into Eqs. (A44) and (A45), we obtain that:
C

4C32 I 2 + C33αβ 2 I + C15α 2

p1V * = , (A89)
8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4
AC

2C34 I 2 + C35αβ 2 I − c2γ 2α 3 β 4

pV2 * = . (A90)
8α (2C22 I 2 + C23 β 2 I ) − γ 2α 3 β 4

After substituting Eqs. (A86)-(A90) into Eqs. (8)-(10), we get that:

I (2C31 I + C12γα 2 β 2 )(4C36 I 2 + 2C37αβ 2 I − C12γα 3 β 4 )
ΠVM*1 = , (A91)
α [8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4 ]2

I 2 (2C38 I + C39αβ 2 ) 2
ΠVM*2 = , (A92)
α [8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4 ]2

41
 ACCEPTED MANUSCRIPT
I 2 (2C38 I + C39αβ 2 ) 2
ΠVM*2 = . (A93)
α [8(2C22 I 2 + C23 β 2 I ) − α 2γ 2 β 4 ]2
Thus, the total profit of the supply chain is as follows:

ΠVSC* = ΠVM*1 + ΠVM*2 + ΠVR* . (A94)

Where

C22 = (α 2 − γ 2 )(2α 2 − γ 2 )2 , C23 = −α (α 2 − γ 2 )(2α 2 − γ 2 ) ,

PT
C24 = (2α 2 − γ 2 )2[aγ (1− ε ) − c1 (α 2 − γ 2 ) + ε aα] ,

C25 = αβ 2[−γ (a − ε a + c1γ )(12α 2 − 5γ 2 ) − 2α(ε a − c1α)(4α 2 − γ 2 ) + c2αγ (4α 2 − 3γ 2 )] ,

RI
C26 = (2α 2 − γ 2 )[c1γ (2α 2 − γ 2 ) − c2α(4α 2 − 3γ 2 ) + a(1− ε )(4α 2 − γ 2 ) + 2ε aαγ ] ,

SC
C27 = [3c1 (2α 2 − γ 2 ) + (aγ − ε aγ + 2ε aα + c2αγ )](α 2 − γ 2 )(2α 2 − γ 2 ) ,

C28 = aαγ (α − εα + εγ ) − (α 2 − γ 2 )[4c1(2α 2 − γ 2 ) + c2αγ ] ,

U
C 29 = (α 2 − γ 2 )(2α 2 − γ 2 )[c1γ (2α 2 − γ 2 ) + c2α (12α 2 − 5γ 2 )
,
+ a (1 − ε )(4α 2 − γ 2 ) + 2ε aαγ ]
AN
C30 = −(α 2 − γ 2 )(2α 2 − γ 2 )(c1αγ + 6c2α 2 ) − aα (1− ε )(2α 2 − γ 2 )2 − ε aγα 2 (2α 2 − γ 2 ) ,
M

C31 = (α 2 − γ 2 )(2α 2 − γ 2 )[aγ (1− ε ) + c2αγ − c1(2α 2 − γ 2 ) + 2aεα] ,

C 32 = a (2α 2 − γ 2 )[2εα (3α 2 − 2γ 2 ) + γ (1 − ε )(5α 2 − 3γ 2 )]
,
D

+(α 2 − γ 2 )(2α 2 − γ 2 )[ c1 (2α 2 − γ 2 ) + c2αγ ]

C33 = (2α 2 − γ 2 )[−5aγ (1− ε ) − 4ε aα − c1(4α 2 − 3γ 2 ) + c2αγ ] ,

TE

C34 = (α 2 − γ 2 )(2α 2 − γ 2 )[c1γ (2α 2 − γ 2 ) + c2α (4α 2 − γ 2 )]

,
+ a (2α 2 − γ 2 )[2εαγ (5α 2 − 3γ 2 ) + (1 − ε )(12α 4 − 9α 2γ 2 + γ 4 )]
EP

C35 = −(2α 2 − γ 2 )[2a(1− ε )(3α 2 − γ 2 ) + c1γ (3α 2 − 2γ 2 ) + c2α(2α 2 − 3γ 2 ) + 3ε aαγ ] ,

C36 = (α 2 − γ 2 )(2α 2 − γ 2 )2[c2αγ − c1 (2α 2 − γ 2 ) + aγ (1− ε ) + 2aαε ] ,

C

C37 = (α 2 − γ 2 )(2α 2 − γ 2 )[ c1 (2α 2 − γ 2 ) − 2c2αγ ]

AC

,
+ a (2α 2 − γ 2 )[(1 − ε )γ 3 − εα (2α 2 − 3γ 2 )]

C38 = (α 2 − γ 2 )(2α 2 − γ 2 )[ c2α (4α 2 − 3γ 2 )

,
− c1γ (2α 2 − γ 2 ) − 2ε aαγ − a (1 − ε )(4α 2 − γ 2 )]

C39 = (γ c1 − 2c2α )(α 2 − γ 2 )(2α 2 − γ 2 ) + a(2α 2 − γ 2 )[(1− ε )(2α 2 − γ 2 ) + εαγ ] ,

42
 ACCEPTED MANUSCRIPT
C40 = a 2 (4α 2 − γ 2 )[(1 − 2ε )(2α 4 − γ 4 ) + 4ε 2α 4 ] + 4αγε (1 − ε )a 2 (2α 2 − γ 2 )(3α 2 − γ 2 )
−2[2ε ac1α + ac1γ (1 − ε ) + αγ c1c2 ](α 2 − γ 2 )(2α 2 − γ 2 ) 2 − 4ε ac2γα 2 (α 2 − γ 2 )(2α 2 − γ 2 )
+c12 (α 2 − γ 2 )(2α 2 − γ 2 )3 + c2α (α 2 − γ 2 )(2α 2 − γ 2 )[c2α (4α 2 − 3γ 2 ) − 2a (1 − ε )(4α 2 − γ 2 )]
−ε 2 a 2γ 2 (α 2 − γ 2 )(2α 2 + γ 2 ) − a 2α 2γ 4 (4ε 2 + 2ε − 1),
C41 = c2 (c1γ − c2α )(a 2 − γ 2 )(2a 2 − γ 2 ) + ε aγ (c2α + c1α
−c1γ )(2a 2 − γ 2 ) + ac2 (1 − ε )(2a 2 − γ 2 ) 2 − a 2 (2a 2 .
−γ )[εγ (1 − ε ) + α ] + aα (2a − γ )[ε a(2 − ε ) − c1γ ]

PT
2 2 2

Proof of Proposition 10. The proof is similar to that of Proposition 2, and is thus omitted
here.

RI
Proof of Proposition 11. Based on the game structure, we first solve for the objective
function of the retailer.

SC
Taking the first order derivative of Eq. (48) with respect to p1 and p2 and setting
them to 0 as follows:

U
∂ΠR
= ε a − α p1 + γ p2 + βθ − ( p1 − ω1 )α + ( p2 − ω2 )γ = 0 , (A95)
∂p1
AN
∂Π R
= (1 − ε )a − α p2 + γ p1 + ( p1 − ω1 )γ − ( p2 − ω2 )α = 0 .
M

(A96)
∂p2
The Hessian matrix of Π R is
D

 ∂ 2Π R ∂ 2Π R 
 
TE

∂p 2 ∂p1 p2   −2α 2γ 
H ( p1 , p2 ) =  2 1 = . (A97)
∂ Π ∂ 2 Π R   2γ −2α 
 R

∂p
 2 1 p ∂p22 
EP

The Hessian matrix of Π R is a negative definite for all values of p1 and p2 if

α > γ . Combining Eqs. (A95) and (A96), we can obtain that:
C

aαε − aγε + ω1α 2 − ω1γ 2 + αβθ + aγ

p1 = , (A98)
2(α 2 − γ 2 )
AC

−aαε + aγε + ω2α 2 − ω2γ 2 + γβθ + aα

p2 = . (A99)
2(α 2 − γ 2 )
Substituting Eqs. (A98) and (A99) into Eqs. (47) and (9), we have
(γω1ω2 − γ c1ω2 − aε c1 + aεω1 + αc1ω1 − αω12 − βθ c1 + βθω1 )
ΠM1 = − Iφθ 2 , (A100)
2
(ω2 − c2 )( a − aε + γω1 − αω2 )
ΠM2 = . (A101)
2

43
 ACCEPTED MANUSCRIPT
Taking the first order derivative of Eq. (A100) with respect to ω1 and θ and the first

order derivative of Eq. (A101) with respect to ω 2 and setting them to 0, we can obtain that:

∂Π M1 γω2 βθ aε − α c1
= −αω1 + + + =0, (A102)
∂ω1 2 2 2
∂Π M 2 γω1 a − aε + α c2
= − αω2 + = 0, (A103)
∂ω2 2 2

PT
∂ΠM1 (ω1 − c1 )β
= − 2Iφθ = 0 . (A104)
∂θ 2

RI
Now, the Hessian matrix is as follows:

 ∂ 2 Π M1 ∂ 2 Π M1   β 
  −α

SC
 ∂ω12 ∂ω1θ   2 
H (ω1 , θ ) =  2 = . (A105)
∂ Π M1 ∂ 2 Π M1   β
   −2 Iφ 
 ∂θω 2 
∂θ   2 


U
1

The Hessian matrix of Π M1 is a negative definite for all values of ω1 and θ if I , φ ,

AN
β2
α and β satisfy the conditions that 2Iφα − >0.
4
M

Solving Eqs. (A102)-(A104), we find that:

2φ IB2 − c1αβ 2
ω1VI * = , (A106)
D

φ IB1 − αβ 2

4φ IB3 − A4 β 2
TE

ω2VI * = , (A107)
2(φ IB1 − αβ 2 )

B4 β
θ VI * =
EP

. (A108)
2(φ IB1 − αβ 2 )
Substituting Eqs. (A106)-(A108) into Eqs. (A98) and (A99), we can derive that:
C

4φ IB5 − B6 β 2
p VI *
= , (A109)
4 A1 (φ IB1 − αβ 2 )
1
AC

4φ IB7 − B8 β 2
p2VI * = . (A110)
4 A1 (φ IB1 − αβ 2 )

Then substituting Eqs. (A106)-(A110) into Eqs. (47), (9) and (48), we can derive that:
φ I (8φ I α − β 2 ) B42
ΠVIM1* = , (A111)
4(φ IB1 − αβ 2 ) 2

α (4φ IB9 − A7 β 2 ) 2
ΠVIM *2 = , (A112)
8(φ IB1 − αβ 2 ) 2

44
 ACCEPTED MANUSCRIPT
4φ Iα 2 B4 (4φ IB10 − A7γβ 2 ) − 4(1 − φ ) IA1 B42 β 2
+α 2 (4φ IB9 − A7 β 2 )(4φ IB11 − A7αβ 2 )
ΠVIR * = . (A113)
16 A1 (φ IB1 − αβ 2 )2
Therefore, the total profit of the supply chain is as follows:

ΠVISC* = ΠVIM1* + ΠVIM2* + ΠVIR *

4φ I (8φ Iα − β 2 ) A1B42 + 2α (4φ IB9 − A7 β 2 )2 A1 − 4(1 − φ ) IA1 B42 β 2

PT
+4α 2φ I (4φ IB10 − A7γβ 2 ) B4 + α 2 (4φ IB9 − A7 β 2 )(4φ IB11 − A7αβ 2 )
= . (A114)
16 A1 (φ IB1 − αβ 2 )2

RI
Proof of Proposition 12. Taking the first derivative of Eq. (51) with respect to ߚ, we get
∂θ VI * (c1γ 2 − 2c1α 2 + 2aαε − aεγ + c2αγ + aγ )(8φ Iα 2 − 2φ I γ 2 + αβ 2 )
= .
∂β 2(8φ Iα 2 − 2φ I γ 2 − αβ 2 )2

SC
β2
Recall that α >γ , 2Iφα − >0 , 8φ I α 2 − 2φ I γ 2 + αβ 2 > 0 , and
4

U
c1γ 2 − 2c1α 2 + 2aαε − aεγ + c2αγ + aγ =aε (2α − γ ) − c1 (2α 2 − γ 2 ) + γ (a + c2α)=B4 > 0 .
AN
∂θ VI *
Then, we have ≥ 0.
∂β
M
D
TE
C EP
AC

45