Research Paper 3

Oper Res Int J
https://doi.org/10.1007/s12351-018-0393-2
ORIGINAL PAPER
Inventory and pricing decisions for a dual‑channel

supply chain with deteriorating products
Yong He1 · Hongfu Huang1 · Dong Li2
Received: 25 December 2016 / Revised: 19 December 2017 / Accepted: 11 March 2018

© Springer-Verlag GmbH Germany, part of Springer Nature 2018
Abstract Dual-channel supply chain structure, i.e., a traditional retail channel

added by an online direct channel, is widely adopted by a lot of firms, including
some companies selling deteriorating products (e.g. fruits, vegetables and meats,
etc.). However, few papers in literature consider deterioration property of products
in dual-channel business models. In this paper, a single-retailer-single-vendor dual-
channel supply chain model is studied, in which the vendor sells deteriorating prod-
ucts through its direct online channel and the indirect retail channel. In addition to
quantity deterioration, quality of the products also drops with time and affects the
demand rate in the retail channel. The pricing decisions and the inventory decisions
for the two firms are simultaneously studied. Models of centralized (i.e., the two
firms make decisions jointly) and decentralized (i.e., the two firms make decisions
separately, vendor as the Stackelberg leader) problems are established. Proper algo-
rithms are proposed to obtain the optimal decisions of prices, ordering frequencies
and ordering quantities. The results suggest that decentralization of the supply chain
not only erodes the two firms’ profit, but also incurs higher wastes comparing to
that under centralization. However, a revenue sharing and two part tariff contract
can coordinate the supply chain. Under utilizing the contract, each firm’s profit is
improved and the total waste rate of the supply chain is reduced. It is also shown
that the contract is more efficient for both firms under higher product deterioration
rate. Besides, the contract is more efficient for the retailer, while less efficient for the
vendor under higher quality dropping rate. In the model extension, online channel
delivery time is assumed to be endogenous and linked to demands in both channels.
The results show that products’ deterioration rate and quality dropping rate have
* Yong He
[email protected]
1
School of Economics and Management, Southeast University, Nanjing 210096, China
2
Management School, University of Liverpool, Liverpool L69 7ZH, UK
13
Y. He et al.
significant impacts to the firms’ delivery time decisions, as well as the pricing and
inventory decisions.
Keywords Dual channel supply chain · Product deterioration · Game theory ·

Pricing · Inventory
1 Introduction
With the fast development of Internet and information technology, customers’ pur-
chasing behaviors have been changed a lot, which pushes more and more firms in
various industries to establish direct selling channels in addition to the traditional
‘Brick-and-Mortar’ channels. By establishing direct channels, companies can often
benefit from the expanded market coverage (Chen et al. 2012), the enhanced con-
trol power over retail price (Chiang et al. 2003), etc. According to a survey, about
42% of the top PC manufacturers (like Dell, Sony, Compaq, Lenovo, etc.) are selling
through their own direct channels (Wilder 1999; Tsay and Agrawal 2004). Today,
in the fresh food industry, with developed preservation and logistic technologies,
many companies are selling their goods to customers through the direct online chan-
nel. For the grocery giant Wal-Mart, fresh foods are sold through Chinese e-com-
merce partner Yihaodian in Shanghai and Beijing.1 Also, Harry and David, as one of
Internet Retailer 500 and America’s leading gourmet gifting companies, sells fresh
foods through both direct and retail channels. Nowadays, in China, more and more
customers choose to buy fruits on e-commerce websites, including Alibaba, Taobao
and Tmall, etc. Khuntonthong et al. (2013) demonstrated that the development of
e-commerce techniques gives farmers more opportunities to benefit from perishable
agricultural foods.
As fresh-selling through dual-channel is becoming more and more popular, the
research on the management of dual-selling for deteriorating products is required
and urgent. However, most of the previous studies on dual-channel problems con-
centrate on the single period pricing problems. Since the product deterioration is
a time linked phenomenon, most of their models cannot characterize product dete-
rioration appropriately. A common way to study product deterioration is to use the
EOQ models.2 Thus, in this paper, in addition to the pricing decisions, inventory
decisions are considered for deterioration items. In the centralized model, the vendor
and the retailer are vertically integrated. They make decisions together on both the
1
Russel, J. (Jul 22, 2015) Walmart Takes Full Control Of Yihaodian, Its Online Retail Business In
China. https://techcrunch.com/2015/07/22/walmart-buys-out-its-chinese-store-yihaodian/. Accessed on
November 7, 2016.
2
The EOQ theory enables people to consider the transaction costs (e.g. deterioration cost, transportation
cost, inventory holding cost, ordering cost) which have great impacts to pricing and ordering decisions
for supply chain members. According to Moss et al. (2003), transaction cost (including deterioration
cost, transportation cost, inventory holding cost, ordering cost, etc.) is an important factor for the applica-
tion of e-commerce business structures. The ignorance of the transaction cost may result in non-optimal
decisions.
13
Inventory and pricing decisions for a dual‑channel supply…
prices of the two channels and the inventory for the whole supply chain. In addition
to the centralized model, a decentralized model is studied, in which, the vendor and
the retailer competes vertically and horizontally on pricing and ordering decisions.
In the game, the vendor is the Stackelberg leader and the retailer is the follower.
Also, a revenue sharing and two part tariff contract is proposed to coordinate the
supply chain, which can not only raise both firms’ profits, but also reduce the total
waste of the whole supply chain. In the model extension, a more realistic model is
studied, in which the direct channel delivery time is endogenous. The main research
questions in this paper can be summarized as:
1. How to obtain the optimal decisions under both centralized and decentralized
supply chains?
2. How does the optimal decisions change with the critical parameters (e.g., dete-
rioration rate, competition intensity, etc.)?
3. Can the revenue sharing and two part tariff contract coordinate the supply chain?
If so, how to determine the optimal contract parameters?
4. How to decide the optimal delivery time when it is endogenous?
The rest of this paper is organized as follows. Section 2 is the literature review. Sec-
tion 3 introduces the notations and assumptions through the paper. Then, the central-
ized and decentralized models are formulated. To solve the models, two algorithms
are also proposed. In Sect. 4, numerical examples and sensitivity analysis are pre-
sented, along with some important results and interesting managerial implications.
In Sect. 5, a revenue sharing and two part tariff contract is proposed to coordinate
the supply chain. Section 6 is the model extension. In the last section, conclusions
for the paper are presented, and some future research topics are suggested.
2 Literature review
This paper is closely linked to two streams: (1) dual channel supply chain models (2)
EOQ/EPQ models with product deterioration.
One stream of literature is about the research of dual-channel business models.
In recent couple of years, dual-channel business model has been deeply studied by
researchers on supply chain management and marketing. Chiang et al. (2003) argued
that the motivation of adding a direct channel is to reduce the double marginaliza-
tion effects. Yan and Pei (2009) pointed out that establishing a direct channel is a
useful tool for the manufacture to motivate the retailer’s service level improvement,
and to enhance the efficiency of the total supply chain. In addition to the pricing
decisions, researchers consider about other important and realistic factors, such as
direct channel delivery lead time, service level for both channels, demand disrup-
tion, asymmetric information, etc. Hua et al. (2010) showed that in a dual-channel
supply, lead time has strong effect on both parties’ pricing and quantity decisions.
Xu et al. (2012) extended Chiang et al. (2003) by treating the delivery time length of
the online channel as a decision variable. They also showed that lead time decision
13
Y. He et al.
has effects to the manufacturer’s channel selections. Yang et al. (2017) studied a
dual-channel Newsvendor model with lead time linked demand and customers’
switching behaviors when shortage occurs. Chen et al. (2017) studied the quality
decisions in addition to the pricing decisions in a dual-channel supply chain. They
showed that adding another channel can improve the product quality and the sup-
ply chain performance. Xiao et al. (2014) investigated the product variety design
and pricing decisions for a two level supply chain in a circular spatial market under
manufacturer-lead and retailer-lead Stackelberg gaming. They found that the motiva-
tion for the manufacturer to use dual channels decreases with the unit production
cost, while increases with the marginal cost of variety design, the retailer’s marginal
selling cost, and the customers’ fit cost. Cai et al. (2009) and Chen et al. (2012) did
excellent study on channel selection policies under different values of selling costs,
potential market share and competition intensity. Dumrongsiri et al. (2008) found
that a higher retail service level or customer service sensitivity can benefit both par-
ties when demand is price and service dependent. Li and Li (2016) studied a dual-
channel supply chain with retailer’s service investment and fairness concerns. Muk-
hopadhyay et al. (2008) studied a dual-channel model with a value adding retailer
who has private information of the value adding costs. They found that a lower cost
for the retailer can induce the information sharing throughout the supply chain.
Dan et al. (2012) argued that the market share and the customers’ loyalty of retail
channel has great influence to the pricing and service decisions. Liu et al. (2015)
studied the manufacturer’s and retailer’s risk aversion behaviors under asymmetric
information to the optimal decisions. Yan et al. (2016) studied the optimal pricing
in a dual-channel supply chain with a dominant retailer and two manufacturers, in
which the manufacturers lie about their cost information. Li et al. (2016a) studied
the pricing and coordination problems in a dual-channel supply chain with a risk-
averse retailer. Liu et al. (2010) demonstrated that when selling cost information is
private, centralized decision is not always better than the decentralized decision with
a feasible contract if the retailer has lower selling cost. Chiang and Monahan (2005)
found that when demand is uncertain and demand can be transferred from one chan-
nel to another, dual-channel is better than either the pure retail channel or the pure
direct channel. Yu et al. (2016) studied the impacts of the e-tailer’s drop-shipping
decisions to the manufacturer’s distribution channel strategies. He et al. (2014) stud-
ied the transshipment strategies between the e-store and the retailer under demand
uncertainties in both channels. Xiao and Shi (2016) studied the optimal pricing
decisions in the presence of supply shortage under different supply priority of each
channel. Yue and Liu (2006) found that for uncertain demand, the gap of demand
estimated by the manufacture and that by the retailer has significant impacts on the
benefit of both parties. Huang et al. (2012) found that when demand is disrupted
and the demand disruption level falls in an interval, the optimal production quantity
decision is robust. Choi et al. (2013) studied the decisions under different power
structures of the dual-channel supply chain when products can be recycled. Mat-
sui (2015) studied the channel strategies (single channel or dual channel) of two
competing manufacturers. They showed that, the symmetry of the two manufactur-
ers can result in asymmetry equilibrium. Matsui (2017) studied the optimal deci-
sion sequence of the direct price and wholesale price in a dual-channel supply chain.
13
They showed that the manufacturer should announce its direct price before or upon
the wholesale price. Lu and Liu (2015) studied the entry of an external e-commerce
channel to the manufacturer’s channel selections. Li et al. (2016b) studied the pric-
ing, greenness and manufacturer’s channel selections in a dual channel green supply
chain. He et al. (2016) studied the carbon emissions in a dual-channel supply chain
in the presence of customers’ free riding behaviors. Ji et al. (2017) studied manufac-
turer’s carbon emission reduction efforts, pricing and channel selection decisions in
a dual-channel supply chain. Takahashi et al. (2011) studied the inventory decisions
in a two echelon dual-channel supply considering the manufacturer’s and retailer’s
stock setup and delivery decisions. Rodriguez and Aydin (2015) studied the pricing
and assortment decisions in a dual-channel supply chain. Chen (2015) studied the
cooperative advertising strategies in a dual-channel supply chain. Xie et al. (2017)
coordinated the dual-channel supply chain in the presence of cooperative advertising
with a revenue sharing contract. Batarfi et al. (2016) studied a centralized decision
model with price competition between the two channels. In addition to the pricing
decisions, they also studied the inventory decisions. Their research is closely related
to this paper. However, this paper is different from Batarfi et al. (2016) in two
aspects. Firstly, product deterioration is considered in this paper, which was seldom
considered in previous research, including Batarfi et al. (2016). Secondly, in Batarfi
et al. (2016), they studied an integrated model. However, this paper also studies a
decentralized model, in which both horizontal and vertical competitions between the
vendor and the retailer are considered.
In addition to the competition problems, researchers also studied the coordina-
tion of the decentralized supply chain with revenue sharing contracts (Yan 2008; Cai
2010; Xie et al. 2017), two-part tariff and profit sharing contracts (Chen et al. 2012),
price discount contracts (Cai et al. 2009) under different situations. In this paper, a
revenue sharing and two part tariff contract is proposed to coordinate the dual chan-
nel supply chain with product deterioration. Summary of the related literature on
dual-channel supply chain is shown in Table 1. To our best knowledge, this paper is
the first that consider product deterioration, inventory decision, supply chain coordi-
nation simultaneously in a dual-channel supply chain.
Another related stream of literature is EOQ/EPQ models for deteriorating prod-
ucts. Summary of the related literature on deteriorating inventory is shown in
Table 2. According to Shah et al. (2013), deterioration is defined as decay, change or
spoilage so that the items are not in its initial conditions. There are two categories of
deterioration items. The first category refers to the items that become decayed, dam-
aged or expired with time, e.g., meat, vegetables, fruits, medicine, etc. The second
category is the items that lose part or total value with time, e.g., computer chips,
mobile phones, fashion and seasonal products, etc. Both kinds of items have short
life cycles and after a period of existence in market, the items lose the original eco-
nomical value due to the drop of consumer preference, product quality, etc. Ghare
and Schrader (1963) first proposed an exponentially decaying inventory model.
Based on their work, people had done a lot on the EOQ problems for deterioration
products. In this research area, different settings of critical factors, e.g., demand rate,
deterioration rate, pricing strategies, etc., have significant impacts on the formula-
tion of the models, and the associated solutions and results. Firstly, for demand rate,
13
Y. He et al.
Table 1 Related literature on dual-channel supply chain

Research paper Pricing Inventory policy SC coordination Others
Chiang et al. (2003)

√
Chen et al. (2012)

√ √
Choi et al. (2013)

√ √
Cai et al. (2009)

√
Chiang and Monahan

√ √
(2005)
Chen (2015) Cooperative advertising
√
Xie et al. (2017) Cooperation advertising

√ √
Yan and Pei (2009) Retailer service

√
Mukhopadhyay et al. Retailer service, information

√
(2008) sharing
Dan et al. (2012) Retailer service
√
Li and Li (2016) Retail service

√
Hua et al. (2010) Lead time

√
Xu et al. (2012) Lead time

√
Yang et al. (2017) Lead time, stochastic demand

√
Xiao et al. (2014) Product variety

√
Dumrongsiri et al. (2008) Stochastic demand

√
Xiao and Shi (2016) Uncertain supply, channel

√
priority
Li et al. (2016a) Risk aversion
√ √
Liu et al. (2015) Risk aversion

√
Yue and Liu (2006) Information sharing

√
Huang et al. (2012) Demand disruption

√
He et al. (2014) Transshipment, stochastic

√
demand
Liu et al. (2010) Stochastic demand, informa-
√
tion sharing
Chen et al. (2017) Quality decision
√
Yu et al. (2016) Drop shipping strategy

√
Matsui (2015) Competition of two manu-

√
facturers
Matsui (2017) Pricing sequence decision
√
Lu and Liu (2015) External e-commerce com-

√
petition
Yan et al. (2016) Retailer dominated SC
√
Li et al. (2016b) Product greenness

√
He et al. (2016) Product greenness, free-

√
riding behavior
Ji et al. (2017) Product greenness
√
Takahashi et al. (2011) Delivery decisions

√
13
Table 1 (continued)
Research paper Pricing Inventory policy SC coordination Others
Rodriguez and Aydin Assortment planning

√ √
(2015)
Batarfi et al. (2016)
√ √ √
This paper Product deterioration

√ √ √
Table 2 Related literature on deteriorating inventory models

Research paper Pricing Deterioration SC level Others
Ghare and Schrader (1963) Quantity One

Sarker et al. (1997) Quantity One Inventory-level dependent
demand
Giri et al. (2003) Quantity One Ramp type demand
Sana et al. (2004) Quantity One Demand shortage
Dye et al. (2006) Quantity One Demand backlogging
Chen and Chen (2007) Quantity Three Multi-item
Lo et al. (2007) Quantity Two Integrated decision
Skouri et al. (2009) Quantity One Weibull deterioration rate
Thangam and Uthayakumar Quantity Two Trade credit
√
(2009)
Lin et al. (2009) Quantity Two Cooperative decision
√
Lin et al. (2010) Quantity Two Cooperative decision

√
Hsu et al. (2010) Quantity One Preservation investment

Liang and Zhou (2011) Quantity One Two warehouse, trade credit
Wang et al. (2011) Quantity Three Time dependent deteriora-
tion rate
Sarkar (2011) Quantity One Trade credit, demand shortage
Mahata (2012) Quantity One Trade credit
Sarkar (2012a) Quantity One Trade credit
Dye and Hsieh (2012) Quantity One Preservation investment
Dye and Hsieh (2013) Quantity One Preservation investment
Sarkar et al. (2013) Quantity One Component cost
√
Sarkar and Sarkar (2013) Quantity One Probabilistic deterioration

Shah et al. (2013) Quantity One Non-instantaneous deteriora-
√
tion
Sarkar (2013) Quantity Two Probabilistic deterioration
Qin et al. (2014) Quantity and quality One Fresh produce
√
Chauhan and Singh (2015) Quantity One Cash flow discount

Sarkar et al. (2015) Quantity One Fixed lifetime
Zhang et al. (2015) Quantity Two Preservation investment
√
This paper Quantity and quality Two Dual channel supply chain
√
13
Y. He et al.
it can either be a constant parameter (Mahata 2012) or be a time dependent parame-

ter (Giri et al. 2003; Wang et al. 2011; Dye et al. 2006; Sarkar 2012a). Also, demand
can be backlogged (Dye et al. 2006), inventory level linked (Burwell et al. 1997;
Sarker et al. 1997) or price sensitive (Shah et al. 2013; Dye and Hsieh 2012; Liang
and Zhou 2011). Secondly, for deterioration rate, it can be a constant parameter
(Sana et al. 2004; Thangam and Uthayakumar 2009; Liang and Zhou 2011; Sarkar
2013; Sarkar et al. 2013), a time linked parameter (Skouri et al. 2009; Sarkar et al.
2015; Sarkar 2011), preservation investment linked parameter (Hsu et al. 2010; Dye
and Hsieh 2013) or a stochastic parameter (Sarkar and Sarkar 2013; Sarkar 2012b,
2013).
The above research only consider the single stage inventory problems. Some peo-
ple studied the problems in multi level supply chains. Lee and Moon (2006) pro-
posed a basic three level producer-vendor-buyer model. Wang et al. (2011) extended
Lee and Moon (2006) by assuming that products suffers from time linked deteriora-
tion rate. Besides, many researchers did a lot of work on integrated inventory and/or
pricing decisions (Lo et al. 2007; Chen and Chen 2007; Noh et al. 2016; Sarkar et al.
2016), Stackelberg gaming problems (Song and Zhao 2010), and cooperation strate-
gies (Lin et al. 2009, 2010; Sarkar 2016) in multi-level supply chains.
To the best of our knowledge, models and analysis in this paper is novel and
different from the extent papers on dual-channel supply chain. The main contribu-
tions of this paper are quadruple. Firstly, product deterioration is considered in a
dual channel supply chain, which is rarely mentioned in previous literature on dual
channel supply chain. Moreover, two kinds of product deterioration is considered,
i.e., the quantity deterioration and the quality deterioration. The quantity deterio-
ration affects the firms’ inventories, whereas the quality deterioration affects cus-
tomers’ choices and demand rate. The problem is modeled over an infinite time
horizon, which enables us to better characterize the products’ deterioration prop-
erty. Secondly, pricing and inventory decisions are studied simultaneously in the
dual channel supply chain, which is seldom considered in previous papers. Thirdly,
endogenous direct channel delivery time is also studied in the dual-channel supply
chain selling deteriorating products. The endogenous direct channel delivery time is
studied by Hua et al. (2010) and Xu et al. (2012). However, how product deteriora-
tion affects the pricing, inventory and delivery time decisions are unclear in their
research. Thus, based on their work, the endogenous delivery lead time problem is
studied in a dual channel supply chain under product deterioration. Lastly, it is found
that a revenue sharing and two part tariff contract can perfectly coordinate the dual
channel supply chain with proper coordinating strategies.
3 Model formulation and solution analysis
The dual-channel supply chain consists of a vendor and a retailer. The supply chain
structure is presented in Fig. 1a, in which the vendor distributes the products to cus-
tomers through both the online channel and the retail channel (Chiang et al. 2003;
Xu et al. 2012). As introduced above, the two parties’ inventory systems are also
considered, which are shown in Fig. 1b. The mathematical models of the problems
13
Fig. 1 Supply chain structure and inventory system
are formulated and algorithms are proposed to solve the models in the following
subsections.
3.1 Notation
Notation in this paper are presented in Table 3.
3.2 Model formulation
Assuming that the vendor is the leader of the supply chain. Firstly, the vendor pro-
cures a large quantity of deteriorating products, noted by Qv. Then the vendor sells
the products to the downstream retailer, as well as to the end customers through
direct channel. Under receiving the ordered products, the retailer sells the products
to the customers. Customers can purchase the products either from the direct chan-
nel or the retail channel. As shown in Fig. 1b, the vendor’s inventory level depletes
due to three reasons: the direct channel demand, the retail channel orders and dete-
riorated quantities. The retailer’s inventory depletes due to retail channel demand
and product deterioration.
Firstly, a centralized model is studied , i.e., the vendor and the retailer are vertically
integrated. There are four decision variables: the retail price pr , the direct channel price
pv, the vendor’s inventory scale parameter n and the retailer’s ordering cycle length
T. Then, a decentralized model is studied, in which the vendor acts as the Stackelberg
leader and the retailer as the follower. In this gaming problem, both of the vendor and
the retailer make their own decisions to maximize their individual profit. The gam-
ing sequence is: (1) The vendor announces the wholesale price w, the direct sale price
13
Y. He et al.
Table 3 Notation
Decision variables
pv Price of the direct channel ($/unit)
pr Price of the retail channel ($/unit)
T Retailer’s replenishment cycle time (day)
n Multiple of retailer’s cycle time (an integer number)
w Wholesale price of the vendor ($/unit)
𝛽 Sharing rate of retailer’s revenue to the vendor
F A lump sum fee transferred from the vendor to the retailer ($)
L Delivery lead time in the direct channel (day)
Constant parameters
𝜃 Deterioration rate of vendor and retailer
𝜇 Quality dropping rate
a Total potential market size
𝛼 Direct channel market share. The retail channel market share is 1 − 𝛼
b Coefficient of the price elasticity of demand rate
r Degree of product substitution of the two channels
cv Vendor’s purchasing cost per unit item ($/unit)
hv , hr Vendor’s and retailer’s holding cost per unit item per unit time, respectively ($/
unit/day)
Av , Ar Vendor’s and retailer’s fixed cost per order, respectively, including order process-
ing cost, transportation cost, warehouse operating cost, etc. ($/time)
𝛾 Vendor’s bargain power. Retailer’s bargain power is 1 − 𝛾 , 0 ≤ 𝛾 ≤ 1
Investment to reduce delivery time, which is formulated as cL = . s and s4 are
s3
cL
L+s4 3
constant parameters ($/time)
Dependent variables
Dv , D r Vendor’s and retailer’s demand rate, respectively
Qv Vendor’s ordering quantity
Qrj Retailer’s ordering quantity in phase j (j = 1, 2, … , n)
Tv Vendor’s replenishment cycle time, Tv = nT
Ivj (t), Irj (t) Vendor’s and retailer’s inventory level with respect to time in the jth phase, respec-
tively, where j = 1, 2, … , n
SRv , SRr Vendor’s and retailer’s total sales revenue, respectively
WRv Vendor’s wholesale total revenue
HCv , HCr Vendor’s and retailer’s total inventory holding cost, respectively
PCv , PCr Vendor’s and retailer’s total purchasing cost, respectively
OCv , OCr Vendor’s and retailer’s ordering cost, respectively
HQv , HQr Vendor’s and retailer’s accumulated holding quantity in a single cycle, respectively
TPv , TPr , TPsc Vendor’s, retailer’s and supply chain’s per unit time profit, respectively
𝛷r , 𝛷v , 𝛷sc Percentage profit increase for retailer ,vendor and supply chain
q ,q
c d Total average ordered quantity for centralized and decentralized model, respec-
tively
D ,D
c d Total average demand for centralized and decentralized model, respectively
B(F) Nash bargain function
13
pv and the scale parameter n. (2) The retailer sets selling price pr and ordering cycle
length T under knowing the vendor’s announced decisions.
Following the studies of Yue and Liu (2006), Huang and Swaminathan (2009), Hua
et al. (2010) and Chen et al. (2012), demand functions of the two channels are linear in
self and cross price affects with the same sensitive parameters. In addition to quantity
deterioration, quality deterioration is also considered in this paper. Following the study
of Wang and Li (2012), Fibich et al. (2003), Kopalle et al. (1996) and Sorger (1988),
we assume that the product quality is exponentially decreasing with time with a rate
of 𝜇. Also, it is realistic that retail channel buyers can touch the products and feel the
quality changes (especially for fresh vegetables, meats and fruits). Thus, the retail chan-
nel demand is linked to the products’ real time quality. Base on the assumptions, the
demand functions in both channels can be formulated as:
Dv (pv , pr ) = 𝛼a − bpv + rpr , (1)
Dr (pv , pr ) = (1 − 𝛼)a − bpr + rpv ]e−𝜇t , (2)

[ ]
t ∈ [0, nT .
In the following analysis, for notational convenience, demand rate of the vendor’s
direct channel is marked by Dv, and that of the retail channel is marked by Dr . Also,
the initial demand rate of the retail channel (i.e., (1 − 𝛼)a − bpr + rpv) is noted as dr .
The demand rates of the two channels are linked to the selling prices in both chan-
nels. The total potential market size is a. 𝛼 (0 < 𝛼 < 1 ) denotes the customer’s pref-
erence for the direct channel, and a higher 𝛼 means more customers will choose the
direct channel. Parameter b is the coefficient of the price elasticity of demand rate.
Parameter r indicates the degree of the substitution of the products sold via the two
channels. To make sure own price effect is greater than cross price effect, the condi-
tion b > r should be satisfied. The parameter 𝜇 denotes the quality dropping rate.
For a higher 𝜇, the product quality drops more fast with time. Specially, when t = 0,
the product quality is 1, which means the product is totally fresh.
To concentrate on the research targets and to ease the analysis, other assumptions
should be made.
1. In the base model, the vendor’s direct channel delivery time is assumed to be zero.
This assumption is relaxed in the model extension by assuming the delivery time
is endogenous and it will affect demands in both channels (Hua et al. 2010).
2. Shortages in both channels are not allowed.
3. Time horizon is infinite for the vendor and the retailer.
4. The market size and deterioration rate do not change with time.
5. No cost is incurred to deal with the deteriorated products. When products are
deteriorated, firms will throw them away without any cost.
3.3 Retailer’s profit
In this subsection, the retailer’s profit is calculated. According to previous studies on

deteriorating inventory problems, based on the inventory system depicted in Fig. 1b,
13
Y. He et al.
some calculations about the inventory level, the ordering quantities, and the total inven-
tory holding quantities for the retailer can be made.
Following Ghare and Schrader (1963), the retailer’s inventory level in phase j satis-
fies the differential equation
İ rj (t) = −𝜃Irj (t) − dr e−𝜇t , t ∈ [(j − 1)T, jT], j = 1, 2, … , n. (3)
with boundary conditions Irj (t = jT) = 0, Irj [t = (j − 1)T] = Qrj . The inventory level
with respect to time t can be derived by solving differential Eq. (3) as
dr
Irj (t) = (e(𝜃−𝜇)jT − e(𝜃−𝜇)t )e−𝜃t , t ∈ [(j − 1)T, jT], j = 1, 2, … , n. (4)
𝜃−𝜇
The ordering quantities of the retailer in phase j can be obtained by equating t to

(j − 1)T , that is
dr
Qrj = Irj [t = (j − 1)T] = (e(𝜃−𝜇)T − 1)e−(j−1)𝜇T , j = 1, 2, … , n. (5)
𝜃−𝜇
The retailer’s total inventory holding quantities in phase j is

jT ( )
dr e𝜃T − 1 e𝜇T − 1 −𝜇jT
HQrj = I (t)dt = − e , j = 1, 2, … , n.
∫(j−1)T rj 𝜃−𝜇 𝜃 𝜇
(6)
After obtaining the inventory level, the ordering quantities, and the total inventory
holding quantities, the related revenue and the costs of the retailer can be obtained
as follows.
(1) Retailer’s total sales revenue (SRr )
The retailer’s total revenue comes from the sales of the deteriorating products. The
total sales revenue in a cycle (t ∈ [0, nT]) can be calculated as
nT
1 − e−𝜇nT
SRr = pr Dr dt = pr dr . (7)
∫0 𝜇
(2) Retailer’s total holding cost ( HCr)

After receiving the ordered products from the vendor, the retailer stores its products
in the inventory. The retailer needs to pay for the holding cost for its inventory. The
total inventory holding cost in an ordering cycle can be calculated as
n ( )
∑ dr e𝜃T − 1 e𝜇T − 1 1 − e−𝜇nT
HCr = hr HQrj = hr − . (8)
j=1
𝜃−𝜇 𝜃 𝜇 e𝜇T − 1
13
(3)Retailer’s total purchasing cost ( PCr)

The retailer purchases its products from the vendor with the wholesale price w.
Thus, the total purchasing cost for the retailer in an ordering cycle can be calculated as
n
∑ dr 1 − e−𝜇nT
PCr = w Qrj = w (e(𝜃−𝜇)T − 1) . (9)
j=1
𝜃−𝜇 1 − e−𝜇T
(4) Retailer’s total ordering cost (OCr)

When ordering from the vendor, the retailer need to pay a lump sum fee Ar per time.
In the cycle t ∈ [0, nT], it will order n times. So the total ordering cost in n phases is
OCr = nAr . (10)
Based on the revenue and cost functions described above, the total profit per unit
time for the retailer can be obtained
1
TPr (pr , T) = [SR − PCr − HCr − OCr ]
nT { r
1 1 − e−𝜇nT
= pr dr
nT 𝜇
dr
( 𝜃T )
e − 1 e𝜇T − 1 1 − e−𝜇nT (11)
− hr −
}
d 1 − e−𝜇nT
−w r (e(𝜃−𝜇)T − 1) − nA r .
The first part of retailer’s profit is the sales revenue. The second part is the retailer’s
total inventory cost. The fourth part is the total purchasing cost. The last part is the
fixed ordering cost.
3.4 Vendor’s profit
In this subsection, the vendor’s profit is calculated. In practice, vendors usually have
inventory holding cost advantage over retailers. Thus, for the vendor, the replenishment
cycle (Tv) is much longer than that of the retailer, which is n times as much as the retail-
er’s ordering cycle (i.e., Tv = nT ). It is more complex for the calculation of the vendor’s
ordering quantity and total inventory holding quantity in time interval t ∈ [0, nT]. There
are n phases in the vendor’s inventory system. For every phase j, ( j = 1, 2, … , n), the
vendor’s inventory level satisfies the following differential equation.
İ vj (t) = −𝜃Ivj (t) − Dv , t ∈ [(j − 1)T, jT], j = 1, 2, … , n. (12)
with boundary conditions (1) Iv(j−1) [t = (j − 1)T] − Ivj [t = (j − 1)T] = Qrj for
j = 2, 3, … , n, (2) Ivn (t = nT) = 0 and (3) Iv1 (t = 0) = Qv − Qr1.
13
Y. He et al.
Lemma 1 The inventory level for the vendor in a cycle in phase j is
Dv 𝜃(nT−t) dr
Ivj (t) = (e − 1) + (e(𝜃−𝜇)nT − e(𝜃−𝜇)jT )e−𝜃t , t ∈ [(j − 1)T, jT]
𝜃 𝜃−𝜇
(13)
in which, j = 1, 2, … , n.
After obtaining the vendor’s inventory level, the total inventory holding quantity in
phase j can be calculated as
jT
Dv e𝜃nT (e𝜃T − 1) −j𝜃T Dv T dr (e𝜃T − 1) (𝜃−𝜇)nT −j𝜃T
HQvj = Ivj (t)dt = e − + (e e − e−j𝜇T ).
∫(j−1)T 𝜃2 𝜃 (𝜃 − 𝜇)𝜃
(14)
The total inventory holding quantity in time interval t ∈ [0, nT] can be calculated as
n ( )
∑ e𝜃nT − 𝜃nT − 1 dr 1 − e−𝜇nT
(15)
(𝜃−𝜇)nT −𝜃nT 𝜃T
HQv = HQvj = Dv + e (1 − e ) + (e − 1) .
j=1 𝜃2 (𝜃 − 𝜇)𝜃 1 − e𝜇T
As shown in Fig. 1b, at time zero, the vendor receives quantity Qv and then transport
Qr1 to the retailer. Then, the rest of the quantity is stocked in the vendor’s warehouse.
According to Eq. (13), the initial inventory level can be calculated as Iv1 (t = 0).
Thus, the vendor’s ordering quantity can be obtained as
Dv 𝜃nT dr
Qv = Iv1 (t = 0) + Qr1 = (e − 1) + (e(𝜃−𝜇)nT − 1). (16)
𝜃 𝜃−𝜇
After obtaining the inventory level, the ordering quantities, and the total inventory
holding quantities, the revenue and cost of the vendor can be expressed as follows.
(1) Vendor’s total sales revenue (SRv )
The vendor’s revenue comes from both channels. It gains revenue by selling to cus-
tomers through direct channel. The total revenue in a cycle can be calculated as
SRv = pv Dv nT. (17)
(2) Vendor’s total wholesale revenue (WRv)
The vendor also gains revenue by wholesaling products to the downstream retailer,
which can be expressed as
n
∑ dr 1 − e−𝜇nT
WRv = w Qrj = w (e(𝜃−𝜇)T − 1) . (18)
j=1
(3) Vendor’s total inventory holding cost ( HCv)

During the selling period, the unsold products are stored in the warehouse.
The vendor needs to pay for holding the inventory. The total inventory holding
cost in a selling cycle can be calculated as
13
( )
e𝜃nT − 𝜃nT − 1 dr (𝜃−𝜇)nT −𝜃nT 𝜃T 1 − e−𝜇nT
HCv = hv HQv = hv Dv + h v e (1 − e ) + (e − 1) .
𝜃2 (𝜃 − 𝜇)𝜃 1 − e𝜇T
(19)
(4) Vendor’s total purchasing cost ( PCv)

Before selling the products, the vendor needs to replenish a large quantity of
products from its upstream suppliers, such as some large farms or food produc-
ers, which a unit replenishment cost of cv . Thus, the total purchasing cost in a
cycle can be calculated as
Dv 𝜃nT d
PCv = cv Qv = cv (e − 1) + cv r (e(𝜃−𝜇)nT − 1). (20)
𝜃 𝜃−𝜇
(5) Vendor’s ordering cost (OCv)

When ordering from the suppliers, the vendor need to pay a lump sum fee,
which is
OCv = Av . (21)
Based on the elements described above, the total profit per unit time for the vendor
is
1
TPv (pv , w, n) = [SRv + WRv − HCv − PCv − OCv ]
nT
1 {
= p D nT
nT v v
d 1 − e−𝜇nT
+ w r (e(𝜃−𝜇)T − 1)
( )
e𝜃nT − 𝜃nT − 1 dr 1 − e−𝜇nT
− hv Dv 2
− hv e(𝜃−𝜇)nT (1 − e−𝜃nT ) + (e𝜃T − 1) 𝜇T
𝜃 (𝜃 − 𝜇)𝜃 1−e
Dv 𝜃nT dr (𝜃−𝜇)nT
− cv (e − 1) − cv (e − 1)
𝜃 𝜃−𝜇
(22)
}
− Av .
The first part of vendor’s profit is the sales revenue from the direct online channels.
The second part is the wholesale revenue from the retail channel. The third part is
the vendor’s total inventory cost in an ordering cycle. The fourth part is the total
purchasing cost for the vendor in a cycle. The last part is the fixed cost in a cycle.
3.5 Analysis of the centralized problem
In this subsection, a centralized dual-channel supply chain is considered, in which

the vendor and the retailer are vertically integrated. In this case, the wholesale price
is only used to divide the total profit between the vendor and the retailer. The deci-
sion variables are pv, pr , n and T. The unit time total profit function of the supply
chain is
13
Y. He et al.
TPsc (pv , pr , n, T) = TPv + TPr

{
1 1 − e−𝜇nT
= pv Dv nT + pr dr
nT 𝜇
( )
e𝜃nT − 𝜃nT − 1 dr (𝜃−𝜇)nT −𝜃nT 𝜃T 1 − e−𝜇nT
− hv Dv − h v e (1 − e ) + (e − 1)
𝜃2 (𝜃 − 𝜇)𝜃 1 − e𝜇T
( 𝜃T 𝜇T
) −𝜇nT
d e −1 e −1 1−e
− hr r −
Dv 𝜃nT dr
− cv (e − 1) − cv (e(𝜃−𝜇)nT − 1)
𝜃 𝜃−𝜇
(23)
}
− nAr − Av .
The problem is to maximize the above function by finding optimal values of pv, pr ,
n and T. Through analysis, the following proposition can be obtained.
Proposition 1 When 𝜃 and 𝜇 are relatively small,

(1) For constant n and T, TPsc is jointly concave in pv and pr.The optimal price in
the direct and retail channel can be respectively expressed as
B2 r + 2bB1 X1 + rX1 B2 − arX1 − arX12 + 𝛼arX12 − 2𝛼abX1 + 𝛼arX1

pc∗
v ≈
,
r2 X12 − 4b2 X1 + 2r2 X1 + r2
(24)
2bB2 + rB1 + rB1 X1 − 2abX1 − 𝛼ar + 2𝛼abX1 − 𝛼arX1

pc∗
r
≈ . (25)
r2 X12 − 4b2 X1 + 2r2 X1 + r2
(2) For constant pv, pr and n, TPsc is concave in T.
√
c∗ 2(nAr + Av )
T ≈ [ ] [ ]
pr dr 𝜇n2 + hv Dv n2 + dr (n − 1)n + hr dr n − cv Dv 𝜃n2 + dr(𝜃 − 𝜇)n2
(26)
(3) For constant pv, pr and T, TPsc is concave in n.
√
2Av
n c∗
≈ (27)
pr dr 𝜇T 2 + hv [Dv T 2 + dr T 2 ] − cv [Dv 𝜃T 2 + dr(𝜃 − 𝜇)T 2 ]
13
Table 4 Algorithm to solve the centralized problem

Algorithm 1
Step 1 Input parameters 𝛼 , a, b, r, hv, hr , cv, Av, Ar , 𝜃 , 𝜇. Set n = 1

Step 2 Set k = 1. Initialize the value of T (k) = 𝛥T ⋅ k in which 𝛥T = 10−2 is the searching step size
Step 3 Substitute
( (k)T (k)and n (k)
(k) into
) Eqs. (24) and (25), and obtain the corresponding values of pv , pr ,
(k) (k)
TPsc pv , pr , n, T
Step 4 Set k = k + 1, then T (k) = 𝛥T ⋅ (k + 1). Go back to Step 3 and obtain the values of p(k+1)
v
, p(k+1)
r ,
TPsc (n, T (k+1) , p(k+1)
v
, p (k+1) )
r
Step 5 Repeat( Step 4, stop until ) k satisfies
( conditions )
TPsc n, T (k) , p(k) (k) ⩾ TP
v , pr sc n, T
(k−1) , p(k−1) , p(k−1)
v r ,
, note k(n) = k , T(n) = T (k), p∗v(n) = p(k)
v ,
( (k) (k) (k)
) ( (k+1) (k+1) (k+1)
) ∗ ∗
TPsc n, T , pv , pr ⩾ TPsc n, T , pv , pr
r and
p∗r(n) = p(k) TP∗sc(n) = TPsc (n, T (k) , p(k)
v
, p(k)
r
)
Step 6 Set n = n + 1, repeat step 2-5 and obtain the corresponding values of k(n+1) ∗
, T(n+1)
∗
, p∗v(n+1),
p∗r(n+1), TP∗sc(n+1)
Step 7
( ) ( )
Repeat Step 6 until n satisfies TP∗sc n, p(∗) , p(∗) , T ∗ ⩾ TP∗sc n + 1, p(∗)
v(n) r(n) (n)
, p(∗) , T ∗
v(n+1) r(n+1) (n+1)
,
( ) ( )
TP∗sc n, p(∗) , p (∗)
v(n) r(n) (n)
, T ∗
⩾ TP∗ n − 1, p(∗)
sc
, p (∗)
,
v(n−1) r(n−1) (n−1)
T ∗
. Output nc∗ = n, T c∗ = T ∗ ,
(nc∗ )
v = pv(nc∗ ), pr = pr(nc∗ ), TPsc = TPsc(nc∗ )

∗ ∗ ∗
pc∗ c∗ c∗
𝜇nT
X1 = 1 − ,
2 ( )
nT (n − 1)T T
(
𝜃nT
) (𝜃 − 𝜇)nT
B1 = −hv b + hv r + hr r − cv b 1 − + cv r 1 − ,
2 2 2 2 2
( )
nT (n − 1)T T
(
𝜃nT
) (𝜃 − 𝜇)nT
B2 = hv r − hv b − hr b + cv r 1 − − cv b 1 − .
2 2 2 2 2
Proposition 1 indicates that TPsc can be maximized for constant n and T. By

updating the values of n and T, the optimal decisions can be found. Thus, a multi-
stages searching method is designed to determine the optimal solutions for the
model. The algorithm is presented in Table 4.
3.6 Analysis of the decentralized problem
In this section, a decentralized supply chain is studied, in which both the vendor and
the retailer make their own decisions to maximize their individual profit. Firstly, the
vendor, as the Stackelberg gaming leader, determines its wholesale price w, the direct
channel price pv and scale parameter n. Then, the retailer, as the follower, sets its sales
price pr and replenishment cycle length T based on the vendor’s decisions. Properties
of the model are listed in Proposition 2 and an algorithm is designed to find the optimal
equilibriums.
13
Y. He et al.
Proposition 2 When 𝜃 and 𝜇 are relatively small,

(1) For constant T and n, the optimal prices pd∗
v
, wd∗ and pd∗
r can be expressed
respectively as
2bC2 + 2rC1 − 2𝛼ab − arX7 + 𝛼ar + 𝛼arX7
pd∗
v ≈
, (28)
2(r2 X7 − 2b2 + r2 )
2b2 C1 − r2 C1 − ab2 X7 + 𝛼ab2 X7 + brX7 C2 − 𝛼abrX7

wd∗ ≈ , (29)
X7 b(r2 X7 − 2b2 + r2 )
1
pd∗
r
≈ ((1 − 𝛼)a + rpd∗
v
+ bhr X4 + bX7 wd∗ ). (30)
2b
(2) For constant price parameters pv, pr , w and n, retailer’s profit function is con-
cave in T.
√
2Ar
T d∗
≈ (31)
pr dr 𝜇n + hr dr + wdr (𝜃 − 𝜇)n
(3) For constant price parameters pv, pr , w and T, vendor’s profit function is con-
cave in n.
√
2Av
nd∗ ≈ (32)
wdr (𝜃 − 𝜇)T 2 + hv [Dv T 2 + dr T 2 ] − cv [Dv 𝜃T 2 + dr(𝜃 − 𝜇)T 2 ]
(𝜃 − 𝜇)nT
X7 = 1 − n,
2
𝜃nT (𝜃−𝜇)nT
hr bT hv rnT hv b(n − 1)T cv r(1 − 2
) cv b(1 − 2
) (𝜃 − 𝜇)nT
C1 = { + − + − }(1 − )n,
4 4 4 2 2 2
(𝜃−𝜇)nT
nT r2 h r(n − 1)T 𝜃nT r2 cv (1 − )r hr rT
C2 = hv ( − b) + v + cv (1 − )( − b) + 2
−
2 2b 4 2 2b 2 4
Proposition 2 indicates that the equilibrium for the decentralized problem exists.
However, due to the complexity, the explicit solutions for parameters n and T can
not be derive. Instead, another algorithm is designed to solve the problem, which is
presented in Table 5.
13
Table 5 Algorithm to solve the decentralized problem

Algorithm 2
Step 1 Input parameters 𝛼 , a, b, r, hv, hr , cv, Av, Ar , 𝜃 , 𝜇. Set n = 1;

Step 2 Set k = 1. Initialize the value of T (k) = 𝛥T ⋅ k , in which 𝛥T = 10−2 is the searching step size;
Step 3 Substitute T (k) and) n into Eqs. (28)-(30)
) and obtain the corresponding values of pv , w ,pr ,
(k) (k) (k)
TPv n, pv , w , and TPr T , pr ;

( (k) (k)
( (k) (k)
Step 4 Set k = k + 1(, then T (k) = 𝛥T) ⋅ (k + 1)(. Go back to Step

) 3 and obtain the values of pv , wv ,
(k+1) (k+1)
p(k+1)
r , TPv n, p(k+1) , w(k+1) and TP T (k+1) , p(k+1) ;
v r r
Step 5 Repeat Step 4, stop until k satisfies conditions TPr T (k) , p(k)
( ) ( )
r
⩾ TPr T (k−1) , p(k−1)
r
,
, note k(n) = k , T(n) = T , pv(n) = pv , pr(n) = p(k) r ,
( (k) (k) ) ( (k+1) (k+1) ) ∗ ∗ (k) ∗ (k) ∗
TPr T , pr ⩾ TPr T , pr
w∗(n) = w(k), TP∗v (n, p(k)
v
, w(k) ) and TP∗r(n) (T (k) , p(k)
r
).
Step 6 Set n = n + 1, repeat(step 2-5 and obtain the ) corresponding ( values of k)(n+1), T(n+1), pv(n+1),
∗ ∗ ∗
pr(n+1), w(n+1), TPv n + 1, pv(n+1) , w(n+1) and TPr(n+1) T(n+1) , pr(n+1) .

∗ ∗ ∗ (∗) (∗) ∗ (∗) (∗)
Step 7
( ) ( )
Repeat Step 6 until n satisfies TP∗v n, p(∗) v(n)
, w(∗)
(n)
⩾ TP∗v n + 1, p(∗)
v(n+1)
, w(∗)
(n+1)
,
( ) ( )
TP∗v n, pv(n) , w(n) ⩾ TP∗v n − 1, pv(n−1) , w(n−1) . Output nd∗ = n, T d∗ = T(n
(∗) (∗) (∗) (∗) ∗
d∗ ),pv
d∗ = p∗
v(nd∗ )
,
pd∗
r
= p∗r(nd∗ ), wd∗ = w∗(nd∗ ), TPd∗
v
= TP∗v(nd∗ ), TPd∗
r
= TP∗r(nd∗ ).
4 Numerical tests
In this section, the proposed models are exemplified by numerical examples,

in which the initial values of the parameters are set as follows: 𝛼 = 0.5, a = 500,
b = 20, r = 5, hv = 0.05$/unit/time, hr = 0.2$/unit/time, cv = 4$/unit, Av = 8000$,
Ar = 100$, 𝜃 = 0.01, 𝜇 = 0.01. Superscripts (⋅)c∗ and (⋅)d∗ are used to denote optimal
decisions under centralized and decentralized problems respectively in the following
analysis.
4.1 Examples for centralized and decentralized models
For the centralized model, using Algorithm 1, the optimal decisions can be deter-
mined: pc∗ v = 10.99
$/unit, pc∗
r = 11.22$/unit, T
c∗
= 2.92day, nc∗ = 10. The maxi-
mum profit is TPsc = 562.34$.
c∗
For the decentralized model, Algorithm 2 is utilized to search for the optimal
decisions. For the vendor, the optimal decision is pd∗ v = 11.11
$/unit, wd∗ = 10.89$/
unit, n = 12, maximum profit is TPv = 458.46$. For the retailer, the optimal deci-
d∗ d∗
sion is pd∗ r = 13.29$/unit, T

d∗
= 2.69day, maximum profit is TPd∗ r = 30.15$. The
total profit of the supply chain is TPd∗sc = 488.62 $, which is less than that of the cen-
tralized supply chain.
( In the centralized ) model, the Total Waste Rate [defined as
1 − Total Ordered Quantity × 100%] is 13.61%. However, in the decentralized model,
Total Demand
the Total Waste Rate rises to 15.09%. Thus, it is worth noting that supply chain
13
Y. He et al.
integration not only results in higher profit, but also helps to reduce the wastes due
to product deterioration.
4.2 Sensitivity analysis on equilibrium strategies
In this subsection, sensitivity analysis is carried out on the equilibrium strategies

with respect to key system parameters 𝜃 , 𝜇, 𝛼 , r, hv, hr , Av, Ar and cv by varying one
parameter once and keeping other parameters fixed.
(1) Sensitivity analysis of deterioration rate 𝜃
Table 6 shows that, in both the centralized and the decentralized models, for a
higher deterioration rate, the direct channel price and the retail channel price increases.
This is because a higher deterioration rate means more products are wasted, thus leads
to higher deterioration cost, especially for the vendor with larger inventory holding
quantities. To protect its profit margin, the firm has strong incentives to enhance both
channels’ prices. Besides, when deterioration rate is higher, the vendor and the retailer
suffers more from product deterioration. To reduce the waste rate and cut down the
deterioration cost, the replenishment cycles of the vendor and the retailer are com-
pressed. In summary, in the two models, for the same value of deterioration rate, the
optimal price for the direct channel under decentralization is slightly higher than that
of the centralized model. Due to the double marginalization effect, the retail channel
price in the decentralized case is much higher than that of the centralized model.
Then, the average quantity and demand change with respect to deterioration rate
is shown in Fig. 2a. In both cases, the average purchased quantity is increasing in
deterioration rate. However, on the contrary, the increased average quantity does not
result in the improvement of market demand. As it is shown, demand rate decreases
dramatically with deterioration rate. The gap between the average quantity and the
demand rate is expending with higher deterioration rate, which means the wasted
quantity of the products is increasing in deterioration rate. Although the size of the
gaps for the decentralized and the centralized cases are of similar sizes, the total
waste rate in the decentralized case is higher than that of the centralized model.
(2) Sensitivity analysis of quality losing rate 𝜇
Under the decentralized scenario, Table 7 shows that the direct channel price
and retail channel price decreases, while the wholesale price increases in 𝜇 . From
economic point of view, when the quality drops faster, the demand in the retail
channel drops and the retailer will set a lower price to stimulate demand. For the
vendor, to get a higher direct channel demand and to mitigate the double mar-
ginalization effect, it will set a lower direct channel price and a higher wholesale
price. The drop of retail channel demand also leads to a lower inventory holding
quantity, so that the two parties’ ordering cycles are shortened. It is natural that
higher quality losing rate leads to the drop of both parties’ profit. Under the cen-
tralized scenario, as 𝜇 increases, pc∗ v
, pc∗
r , Tv (T
d∗ d∗
⋅ nd∗ ) and TPc∗
sc
drops, which in
line with the results in the decentralized scenario.
Then, the average quantity and demand change with respect to quality los-
ing rate 𝜇 is shown in Fig. 2b. In both cases, the average purchased quantity and
13
Table 6 Optimal centralized and decentralized decisions for different 𝜃
Centralized supply chain Decentralized supply chain
pc∗
v
pc∗
r T c∗ nc∗ TPc∗ pd∗
v
wd∗ pd∗
r
T d∗ nd∗ TPd∗
v
TPd∗
r
TPd∗
0 10.72 11.04 3.50 11 690.25 10.83 10.82 13.16 2.94 15 581.05 33.87 614.92
0.005 10.86 11.14 3.30 10 624.77 10.98 10.85 13.23 2.85 13 516.93 32.82 549.75
𝜃 0.010 10.99 11.22 2.92 10 562.34 11.11 10.89 13.29 2.69 12 458.46 30.15 488.62
0.015 11.09 11.30 2.90 9 505.95 11.21 10.91 13.34 2.59 11 404.53 28.13 432.66
0.020 11.19 11.38 2.65 9 450.68 11.34 10.98 13.42 2.41 11 353.82 23.27 377.09
13
13
Table 7 Optimal centralized and decentralized decisions for different 𝜇
pc∗
v
pc∗
v
wd∗ pd∗
r
T d∗ nd∗ TPd∗
v
TPd∗
r
TPd∗
0 11.09 11.21 3.04 10 617.12 11.18 10.82 13.30 3.04 11 485.80 47.16 532.96
0.005 11.03 11.21 2.97 10 588.39 11.11 10.82 13.29 2.89 11 471.41 39.68 511.08
𝜇 0.010 10.99 11.22 2.92 10 562.34 11.11 10.89 13.29 2.69 12 458.46 30.15 488.62
0.015 10.94 11.24 3.17 9 540.89 11.07 10.91 13.28 2.63 12 446.46 24.05 470.51
0.020 10.91 11.27 3.14 9 519.68 11.04 10.94 13.27 2.58 12 435.46 18.56 454.02
Y. He et al.
200 c 200 c d
qc D qd D
d
qc D qd D
180 180
160 160
140 140
120 120
0 0.005 0.01 0.015 0.02 0 0.005 0.01 0.015 0.02
θ µ
(a) Average quantity and demand change with θ (b) Average quantity and demand change with µ
190 c d
190 c d
qc D qd D qc D qd D
180 180
170 170
160 160
150 150
140 140
130 130
120 120
0.45 0.5 0.55 2 4 6 8
α r
(c) Average quantity and demand change with α (d) Average quantity and demand change with r
Fig. 2 Average quantity and demand change with respect to 𝜃 , 𝜇, 𝛼 , and r, respectively
total average demand decline in 𝜇 . As it is shown, the average purchased quantity

and total average demand drops in similar speed. The gap between the average
quantity and the demand rate is shrinking with higher 𝜇 , which means the wasted
quantity of the products is decreasing in 𝜇 . Although the size of the gaps for the
decentralized and the centralized cases are of similar sizes, the total waste rate in
the decentralized case is higher than that of the centralized model.
(3) Sensitivity analysis for market share 𝛼 and competition intensity r
Table 8 shows that, when the direct channel market share (𝛼 ) increases, for
both the centralized and decentralized models, the direct channel price increases
while the retail channel price decreases. The wholesale price drops with 𝛼 in the
decentralized case. The replenishment cycle for the retailer increases in 𝛼 while
vendor’s ordering cycle decreases in both cases. For the decentralized model,
when direct channel market share 𝛼 increases, the vendor has more power in the
13
Y. He et al.
market, so it can set a higher selling price. However, for the retailer, to survive
on the market, it would set a lower selling price. In order to balance the revenue
of the two channels, the vendor transfers some demand to the retailer by setting
a lower wholesale price, so that the retailer can set a lower price and increase the
retailer channel demand. The market share of direct channel contributes to the
profit of the total supply chain profit under both cases and vendor’s profit under
decentralized case. However, the retailer’s profit is hurt due to the drop of its mar-
ket power. In Fig. 2c, it is depicted that the average quantity and demand are not
sensitive to 𝛼 in the centralized case. However, under the decentralized case, the
average quantity and demand both increase with 𝛼 , the gap remains unchanged,
which means the direct channel power contributes to the improvement of demand
and to the reduction of total waste.
Next, the influence of competition intensity to the optimal decisions are studied.
Based on Chen et al. (2012), the competition intensity can be denoted as r, where
r = b − b ( b is a constant). Note that, when r = 0, there is no competition between
� ′
the two channels. In this paper, b is set as b = 15, and the corresponding sets of
′ �
r and b are (r, b) = {(1, 16), (3, 18), (5, 20), (7, 22), (9, 24)}. The result is presented
in Table 8. It is shown that when the competition intensity increases, for both the
centralized and decentralized models, the direct channel price decreases. The retail
channel price increases in the centralized model, while decreases in the decentral-
ized model. The ordering cycle of centralized model and the decentralized model
are increasing in the competition intensity. It shows in Fig. 2d that the average
ordering quantity and profit of centralized model are slightly sensitive to compe-
tition intensity. When competition intensity increases, an integrated vendor should
increase the price gap between the direct channel and the retail channel to keep the
demand, ordering quantity and profit stable. It also shows in Fig. 2d that both the
average quantity and the demand increase with competition intensity in the decen-
tralized case. For the decentralized model, the intensified competition will push the
retailer to set a lower selling price. In response, to maintain the demand in both
channels, the vendor will also set lower wholesale and direct selling price. This pric-
ing strategy helps to mitigate the double marginalization effect between the two par-
ties, thus results in a significant enhancement of total supply chain profit. However,
when the vendor acts as the game leader, competition intensity only has positive
effect on the vendors profit, while has negative effect on the retailers profit. In sum-
mary, in the centralized model, competition intensity has minor effects to the prof-
itability of the supply chain. In the decentralized model, intense competition will
benefit the vendor, while will hurt the retailer. Numerically, Fig. 2d also tells that the
gap between average quantity and demand is not sensitive to competition intensity,
while the waste rate drops under higher competition intensity.
(4) Sensitivity analysis for cost parameters hv, hr , Av, Ar , cv
We show the sensitivity results of hv, hr , Av, Ar , and cv in Table 9. In the decen-
tralized case, when the vendor’s inventory holding cost hv increases, pd∗ v
, pd∗
r and w
d∗
increase, whereas TPv , TPr and TPsc decrease. A larger inventory holding cost hv
d∗ d∗ d∗
pushes the vendor to set higher selling and wholesale prices so as to protect its sales
margin. The retailer will also increase its selling price under a higher wholesale
13
Table 8 Optimal centralized and decentralized decisions for different 𝛼 and r
pc∗
v
pc∗
v
wd∗ pd∗
r
T d∗ nd∗ TPd∗
v
TPd∗
r
TPd∗
0.450 10.46 11.69 2.85 10 557.67 10.60 11.42 14.09 2.30 14 407.37 47.82 455.19
0.475 10.72 11.45 2.89 10 557.05 10.86 11.16 13.69 2.49 13 430.84 38.56 469.39
𝛼 0.500 10.99 11.22 2.92 10 562.34 11.11 10.89 13.29 2.69 12 458.46 30.15 488.62
0.525 11.25 10.99 3.25 9 575.36 11.37 10.63 12.90 2.98 11 489.84 22.61 512.45
0.550 11.51 10.76 3.28 9 592.53 11.62 10.35 12.50 3.26 10 525.66 15.93 541.60
1 11.04 11.16 2.91 10 561.83 11.13 10.96 13.77 2.30 14 416.36 45.77 462.14
3 11.01 11.19 2.91 10 562.05 11.12 10.92 13.51 2.49 13 439.79 36.78 476.57
r 5 10.99 11.22 2.92 10 562.34 11.11 10.89 13.29 2.69 12 458.46 30.15 488.62
7 10.97 11.24 3.22 9 564.41 11.09 10.87 13.12 2.92 11 473.67 25.23 498.90
9 10.96 11.26 3.22 9 564.77 11.06 10.85 12.95 3.11 10 486.53 21.58 508.11
13
Y. He et al.
price. The increase of prices results in the drop of market demand, thus leads to
the drop of profit in both channels and the total supply chain. Although the dropped
demand rate results in lower total inventory holding cost, it can not compensate the
loss of demand decline. A similar rationale can be applied to the centralized case, in
which prices for the two channels pc∗ v
and pc∗
r increase and total supply chain profit
c∗ decreases with a higher inventory holding cost h .
TPsc v
For a higher value of hr, in the decentralized case, the vendor’s wholesale price wd∗
decrease, while the retailer’s selling price pd∗r increases and ordering cycle T decreases.
d∗
It is common that the retailer will enhance its market price and cut the ordering cycle to
obtain higher sales margin and reduce its total inventory holding cost, although it leads
to the drop of retailer’s ordering quantity. Thus, to stimulate the retailer to order more
products, the vendor sets a lower wholesale price. In the centralized case, higher inven-
tory holding cost in the retail channel leads to the rise of both channels’ prices.
For a higher value of parameter Av, in the decentralized case, prices in the two
channels ( pd∗ v
, pd∗
r and w ) and ordering cycle of the vendor Tv (T
d∗ d∗ d∗
⋅ nd∗ ) increase,
while the retailer’s ordering cycle T decreases. To reduce the high ordering cost,
d∗
the vendor sets a longer ordering cycle. However, longer ordering cycle means larger
ordering quantity and inventory holding cost. To balance the fixed ordering cost and
total inventory holding cost, the vendor also sets higher prices to keep a low demand
rate, thus it can achieve smaller ordering quantity, lower inventory level and lower
inventory holding cost. No doubt that the increase of ordering cost leads to the drop
of both firms’ profit. A simple rationale can be applied to explain why prices and
profit drops in the centralized case.
As the retailer’s ordering cost Ar increase, under the decentralized scenario, the
retailer’s ordering cycle T d∗ increase without doubt. The vendor’s ordering cycle
Tvd∗ (T d∗ ⋅ nd∗) increases in Ar . The prices pd∗v
, pd∗
r in the two channels increase, while
d∗ decreases in parameter
w Ar . This is because for a higher Ar , the retailer’s ordering
frequency nd∗ drops, which leads to the rise of the vendor’s inventory holding costs.
To encourage the retailer to order more products, the vendor offers a lower whole-
sale price. All these leads to the drop of the total profit, vendor’s and retailer’s profit
under higher value of Ar . In the centralized case, the Ar is an internal operating cost,
which has small impacts to the pricing decisions, but has negative impacts to the
total profit.
For a higher procurement cost cv, in the decentralized and the centralized case,
the prices pd∗ v
, pd∗
r , w and ordering cycle increase T , while total profit TPsc , ven-
d∗ d∗ d∗
dor’s profit TPv and retailer’s TPr profit drop. In the centralized case, prices pc∗
d∗ d∗
v
, pc∗
r ,
c∗, ordering cycle T c∗ increases, while the total profit
w TPsc decreases in cv. It is
c∗
because, when procurement cost increases, firms need to set higher selling prices,
which leads to the drop of market demand. Also, to reduce its inventory cost, the
vendor should set a longer ordering cycle.
5 Supply chain coordination
In this section, a revenue sharing and two part tariff contract is utilized to coordi-
nate the supply chain. When adopting the contract, the retailer commits to share a
13
Table 9 Optimal centralized and decentralized decisions for different hv, hr , Av, Ar and cv
pc∗
v
pc∗
v
wd∗ pd∗
r
T d∗ nd∗ TPd∗
v
TPd∗
r
TPd∗
Default 10.99 11.22 2.92 10 562.34 11.11 10.89 13.29 2.69 12 458.46 30.15 488.62
− 40% 10.88 11.14 2.90 11 610.41 10.99 10.81 13.24 2.73 13 505.14 31.37 536.52
hv − 20% 10.93 11.18 3.03 10 586.47 11.03 10.82 13.25 2.73 12 481.02 32.01 513.02
20% 11.03 11.26 3.11 9 541.31 11.130 10.89 13.30 2.73 11 436.82 31.30 468.12
40% 11.08 11.29 3.01 9 519.47 11.20 10.95 13.33 2.68 11 416.60 29.62 446.22
− 40% 10.98 11.17 4.12 7 577.79 11.13 10.93 13.25 2.84 11 461.95 34.97 496.92
hr − 20% 10.98 11.20 3.23 9 568.61 11.12 10.90 13.29 2.73 12 460.07 31.08 491.15
20% 10.99 11.24 2.90 10 558.27 11.10 10.86 13.30 2.66 12 456.77 29.24 486.00
40% 11.00 11.26 2.65 11 552.90 11.09 10.83 13.31 2.65 12 455.00 28.34 483.34
− 40% 10.82 11.03 2.81 8 690.56 10.91 10.64 13.17 3.09 8 570.93 42.61 613.54
Av − 20% 10.91 11.13 2.89 9 622.34 11.01 10.77 13.231 2.86 10 511.13 36.03 547.16
20% 11.06 11.31 3.19 10 509.89 11.20 11.00 13.35 2.56 14 410.85 24.82 435.67
40% 11.13 11.38 3.15 11 460.26 11.26 11.06 13.39 2.52 15 367.21 22.33 389.55
− 40% 10.97 11.19 2.60 11 575.50 11.09 10.99 13.26 1.67 19 461.14 33.29 494.43
Ar − 20% 10.98 11.21 2.88 10 569.24 11.08 10.91 13.26 2.23 14 459.59 33.15 492.75
20% 10.99 11.23 3.25 9 557.89 11.10 10.82 13.30 3.21 10 457.16 29.46 486.62
40% 11.00 11.24 3.28 9 551.76 11.11 10.79 13.31 3.62 9 456.01 27.16 483.17
− 40% 10.03 10.30 2.86 10 875.83 10.13 9.99 12.69 2.41 13 721.70 52.44 774.14
cv − 20% 10.51 10.75 2.88 10 713.06 10.64 10.47 13.01 2.52 13 584.76 39.38 624.14
20% 11.47 11.69 2.97 10 423.78 11.57 11.31 13.58 2.91 11 342.26 21.92 364.18
40% 11.96 12.18 3.36 9 299.38 12.12 11.80 13.91 3.10 11 236.37 11.96 248.33
13
Y. He et al.
proportion of 𝛽 of its revenue with the vendor and the vendor sets a lower wholesale
price w. Then, the two firms negotiate on the lump sum fee F based on their powers.
Use superscript (⋅)co to denote the parameters under supply chain coordination.
5.1 Supply chain coordination with a revenue sharing and two part tariff
contract
The unit time total profits of the retailer and the vendor are respectively given as
1
TPco
r
= [(1 − 𝛽)SRr − PCr − HCr − OCr ]
nT { ( 𝜃T )
1 1 − e−𝜇nT dr e − 1 e𝜇T − 1 1 − e−𝜇nT
= (1 − 𝛽)pr dr − hr −
nT 𝜇 𝜃−𝜇 𝜃 𝜇 e𝜇T − 1
}
d 1 − e−𝜇nT
− w r (e(𝜃−𝜇)T − 1) − nAr + F.
(33)
1
TPco = [𝛽SR + SRv + WRv − HCv − PCv − OCv ]
v nT { r
1 1 − e−𝜇nT d 1 − e−𝜇nT
= 𝛽pr dr + pv Dv nT + w r (e(𝜃−𝜇)T − 1)
nT 𝜇 𝜃−𝜇 1 − e−𝜇T
( )
𝜃nT
e − 𝜃nT − 1 dr (𝜃−𝜇)nT −𝜃nT 𝜃T 1 − e−𝜇nT
− hv Dv − h v e (e − 1) + (e − 1)
𝜃2 (𝜃 − 𝜇)𝜃 e𝜇T − 1
}
D d
− cv v (e𝜃nT − 1) − cv r (e(𝜃−𝜇)nT − 1) − Av − F.
𝜃 𝜃−𝜇
(34)
In the above two functions, it is shown that when F is negative, the lump sum fee is
transferred from the retailer to the vendor. The following proposition characterizes
the conditions that a revenue sharing and two part tariff contract can coordinate the
decentralized supply chain.
Proposition 3 When 𝜃 and 𝜇 are relatively small, the supply chain coordination
can be achieved through a revenue sharing and two part tariff contract only if the
mechanism (𝛽 co, wco, F) satisfies
b(2 − (𝜃 − 𝜇)nc T c )(hr drc − 2Ar ∕T c2 ) + drc hr (𝜃 − 𝜇)bnc T c

𝛽 co = 1 − ∈ [0, 1],
(drc − bpcr )drc (𝜇nc T c − 2)(𝜃 − 𝜇)nc + pcr drc b((𝜃 − 𝜇)nc T c − 2)𝜇nc
(35)
(drc − bpcr )(2 − 𝜇nc T c )(2Ar ∕T c2 − hr drc ) + pcr drc bhr T c 𝜇nc
wco = > 0,
(drc − bpcr )drc (𝜇nc T c − 2)(𝜃 − 𝜇)nc + pcr drc b((𝜃 − 𝜇)nc T c − 2)𝜇nc
(36)
13
F ∈ {TPco
r
(𝛽 co , wco , F) ⩾ TPd∗
r
, TPco
v
v
}, (37)
in which drc
= (1 − 𝛼)a − bpcr + rpcv, pcv , pcr , nc , T c are the optimal decisions in the
centralized model.
According to the contract structure, three coordinating tools can be used by both
players to establish a efficient solution, i.e., the wholesale price, the revenue shar-
ing rate, and the lump sum fee. In this contract, F co is in the range [F, F], which
allows the retailer to earn no less profit than that available in the decentralized
model (TPd∗r ). Consider TPr as the retailer’s reservation profit-the lowest profit level
d∗
at which the retailer can accept the contract. Eqs. (33) and (34) show that higher F co
benefits the retailer, whereas lower F co benefits the vendor. Negotiating the value of
F co depends heavily on the bargain power of the retailer and the vendor in the supply
chain.
5.2 The optimal lump sum fee
Nash bargain method can be applied to determine the optimal F, which can achieve
the Pareto improvement. According to Baron et al. (2016), Nash bargain function
B(F) is modeled as follows
B(F) = [TPco
v
(𝛽 co , wco , F) − TPd∗
v
]𝛾 [TPco
r
(𝛽 co , wco , F) − TPd∗
r
]1−𝛾 . (38)
The goal of the negotiation is to maximize Nash bargain function B(F) by finding
an optimal F. In the function, parameter 𝛾, (𝛾 ∈ [0, 1]) denotes the vendor’s bargain
power. The optimal F can be obtained by solving the equation 𝜕B(F) 𝜕F
= 0, which is
shown in Proposition 4.
Proposition 4 For fixed 𝛾 , the optimal lump sum fee is
F co = [TPco
v
(𝛽 co , wco , F = 0) − TPd∗
v
](1 − 𝛾) − [TPco
r
(𝛽 co , wco , F = 0) − TPd∗
r
]𝛾.
(39)
Proposition 4 shows that the optimal lump sum fee is decreasing in the vendor’s
bargain power.
5.3 An example
The objective of this subsection is to gain further insights of the coordination strat-
egy through numerical tests. The parameters in the base model are also used in this
subsection. Substituting the parameters into Eqs. (35) and (36), the coordinating rev-
enue sharing rate can be obtained as 𝛽 co = 0.698, the wholesale price is wco = 1.86
$/unit. The profit of the retailer under coordination is TPco r = 50.82$ + F , and the
vendor’s profit is TPco
v = 511.52$ − F . The profit of the vendor, the retailer and the
supply chain with and without coordination are plotted in Fig. 3 with respect to F.
Considering the condition in Eq. (37), by choosing a proper value of F in the range
13
Y. He et al.
600 co
T Psc
d
T Psc
500
T Pvd
400
T Pvco
300 P areto Zone

T Prco
200
100 F = −20.666 F = 53.058

T Prd
0
-50 0 50 100 150 200
F
Fig. 3 Change of vendor, retailer and supply chain’s profit with respect to F before and after coordina-
tion
F ∈ [−20.66, 53.06], the coordinated supply chain can reach Pareto improving. For a
fixed 𝛾 , the optimal lump sum fee is F co = 50.06 − 73.72𝛾 . In this example, the neg-
ativity of F co means the lump sum fee is transferred from the retailer to the vendor.
The impacts of parameters 𝜃 , 𝜇 on the efficiency of the coordinating contract
TPco −TPd
under fixed 𝛾 are studied. 𝛷i = iTPd i × 100% denotes the Percentage Profit
i
Increase of i, (i = v, r, sc), which also captures the coordination efficiency of the

contract. Fig. 4a, b depict 𝛷i with respect to parameters 𝜃 and 𝜇.
It is shown in Fig. 4a, for the vendor, the retailer and the supply chain, the coordi-
nation efficiency increases in the product deteriorating rate 𝜃 . It means that, supply
chain members are more willing to coordinate with each other when deterioration
rate is high. Comparing 𝛷v and 𝛷r , the retailer will benefit more from coordination.
It is depicted in Fig. 4b that when the quality of the product drops more fast, the
coordination efficiency for the retailer rises rapidly. However, the coordination effi-
ciency of the vendor drops slightly. In another word, when quality drops more fast,
the retailer is more willing to coordinate with the vendor. But quality dropping rate
has no significant impacts to the vendor’s coordinating efficiency.
6 Extension: delivery time as a decision variable
In this section, the direct channel delivery time, i.e., L, is assumed to be an endog-
enous parameter. In real life, when the direct channel delivery time is longer, more
customers will switch from the direct channel to the retail channel. Following Hua
et al. (2010), it is assumed that the direct channel demand is decreasing in L, while
the retail channel demand is increasing in L. Then, the demand functions for the two
channels can be formulated as
13
35 Φr Φv Φsc
30 Φr Φv Φsc
Percentage Profit Increase
Percentage Profit Change

30
25
25
20
20
15
15
10 10
0 0.005 0.01 0.015 0.02 0 0.005 0.01 0.015 0.02
θ µ
(a) Percentage profit increase with θ (b) Percentage profit change with µ
Fig. 4 Percentage profit increase with respect to 𝜃 or 𝜇 when 𝛾 = 0.9
DLv = 𝛼a − bpv + rpr − s1 L, (40)
DLr = [(1 − 𝛼)a − bpr + rpv + s2 L]e−𝜇t , t ∈ [0, nT]. (41)

in which, s1 and s2 are the lead time sensitivity of the demands in the direct and retail
channel, respectively. Here, the delivery time L is controllable, which incurs an
investment cost cL (L). Assuming the cost cL (L) is decreasing and convex in L, which
follows the law of Diminishing Marginal Utility. Similar to Hua et al. (2010), the
cost is formulated as cL (L) = L+s3 . Note [(1 − 𝛼)a − bpr + rpv + s2 L] as drL in
s
4
Eq. (41). Under the consideration of endogenous delivery time, the two firms’ profit
can be formulated as
{
dL
( )
1 1 − e−𝜇nT e𝜃T − 1 e𝜇T − 1 1 − e−𝜇nT
TPLr (pr , T) = pr drL − hr r −
nT 𝜇 𝜃−𝜇 𝜃 𝜇 e𝜇T − 1
}
drL 1 − e−𝜇nT
−w (e(𝜃−𝜇)T − 1) − nAr .
(42)
{
1 drL 1 − e−𝜇nT
TPLv (pv , w, n, L) = pv DLv nT + w (e(𝜃−𝜇)T − 1)
nT 𝜃−𝜇 1 − e−𝜇T
drL
( )
e𝜃nT
− 𝜃nT − 1 (𝜃−𝜇)nT −𝜃nT 1 − e−𝜇nT
− hv DLv − h v e (1 − e ) + (e 𝜃T
− 1)
𝜃2 (𝜃 − 𝜇)𝜃 1 − e𝜇T
}
D L d L
− cv v (e𝜃nT − 1) − cv r (e(𝜃−𝜇)nT − 1) − Av − cL (L).
𝜃 𝜃−𝜇 (43)
In the centralized model, the two firms make decision together to maximize the
total profit by deciding optimal pv, pr , T, n and L. In the decentralized model, as
the leader, the vendor first determines pv, w, n and L; then, the retailer sets pr and T
13
Y. He et al.
Centralized Decision
490
Supply Chain's profit

485
480
475
470 Lc∗ = 0.86
465
0 0.5 1 1.5 2
L
Decentralized Decision
375
370
Vendor's Profit
365
360
355 Ld∗ = 0.61
350
0 0.5 1 1.5 2
L
Fig. 5 Profit change with respect to L in the centralized and decentralized supply chain
optimally based on the vendor’s strategies. Considering endogenous direct channel

delivery time makes the model more complex and it is hard to obtain the analytical
results. Numerical tests are conducted to find some important managerial insights.
Values for parameters of s1, s2, s3 and s4 are set as 10, 5, 100 and 1, respectively.
Other parameters are set as that in the base model. For constant L, the optimal pric-
ing decisions and inventory decisions in the centralized and decentralized models
can be obtained by utilizing Algorithm 1 and 2, respectively. Then, with iteration of
parameter L, the optimal delivery time can be found.
It is shown in Fig. 5 that, in the centralized supply chain, the total profit is con-
cave in L and the optimal point is Lc∗ = 0.86 day. Then, in the decentralized model,
the vendor’s profit is also concave in L and the optimal point is Ld∗ = 0.61 day. It
is found that, in the decentralized supply chain, the vendor should invest more to
shorten the delivery time comparing to the centralized model, which conforms to the
conclusions in Hua et al. (2010).
The sensitivity analysis of optimal delivery time in the centralized and decentral-
ized models with respect to 𝜃 and 𝜇 are also conducted, which are shown in Fig. 6a,
b. When deterioration rate (𝜃 ) rises, the vendor will invest less in reducing the deliv-
ery time in both the decentralized and centralized models. This is because shorter
delivery time contributes to the total demand rate, while, it also results in the dete-
rioration of more products. Thus, to avoid the deterioration cost, the vendor has less
incentives to reduce the delivery lead time when 𝜃 is higher. When the quality drop-
ping rate (𝜇) is higher, in both models, vendor will invest more to achieve a shorter
delivery time. This is because higher 𝜇 results in quicker shrink of retail channel
13
Ld∗ Ld∗
0.9 0.9 Lc∗
Lc∗
Optimal Delivery Time
Optimal Delivery Time

0.85 0.85
0.8 0.8
0.75 0.75
0.7 0.7
0.65 0.65
0.6 0.6
0.55 0.55
0 0.005 0.01 0.015 0.02 0 0.005 0.01 0.015 0.02
θ µ
(a) Optimal delivery time change with θ (b) Optimal delivery time change with µ
Fig. 6 Optimal delivery time in centralized and decentralized models with respect to 𝜃 or 𝜇
demand. To decelerate the demand dropping, the vendor should invest more on the
delivery time and attract more customers.
7 Conclusions
Nowadays, dual-channel supply chain structure, i.e., a traditional retailer channel

and an online direct channel, is widely adopted by some companies selling deterio-
rating products. However, few literature considers about the deterioration property
of products in dual-channel supply chain decisions. To fill this gap, centralized and
decentralized models of a dual-channel supply chain for deterioration products with
a vendor and a retailer are established. The two firms make decisions on pricing and
inventory under channel competition and product deterioration. In the centralized
model, the vendor and the retailer make decisions together to maximize the total
profit of the supply chain. In the decentralized model, the vendor and retailer com-
petes with each other and follows a Stackelberg gaming sequence in which the ven-
dor acts as the leader. Due to the complexity of the models, no explicit solution can
be obtained by calculation. Thus, two algorithms are proposed to solve the models.
The key findings are obtained through model analysis and numerical tests. The
impacts of critical parameters (deterioration rate, competition intensity, inven-
tory holding cost, ordering cost, etc.) to the price and inventory decisions are
presented, from which, some interesting and meaningful results are summarized.
In addition, it is found that a revenue sharing and two part tariff contract can
coordinate the supply chain. Under applying the coordination contract, the total
waste rate of the supply chain declines. In the extension, the optimal direct chan-
nel delivery time is studied. Results show that the vendor will invest more in
reducing the delivery lead time in the decentralized model than in the centralized
model. In addition, in both models, vendor will set a higher delivery lead time
when deterioration rate is higher or quality dropping rate is smaller.
13
Y. He et al.
There are still some limitations of this paper. Firstly, in this paper, the deterio-
ration rate does not change with time. However, in some cases, deterioration rate
is not a constant, and it often changes with time. For example, Qin et al. (2014)
and Chauhan and Singh (2015) assumed that the deterioration rate is increasing in
time. Skouri et al. (2009) studied an inventory model with Weibull deterioration
rate. Shah et al. (2013) studied a model with non-instantaneously deteriorating
products with time dependent deterioration rate. In the future research, this study
can be extended by incorporating time dependent deterioration rate. Secondly, the
deterioration cost is not considered in this paper. However, in some conditions,
companies also need to pay when dealing with the wastes (Zhang et al. 2015;
Qin et al. 2014). In the future, deteriorating cost will be introduced to the model.
Thirdly, it is assumed that only retail channel customers are sensitive to the prod-
ucts’ quality because they can touch and feel the quality before their purchase.
This assumption can also be relaxed by assuming both channel’s customers are
sensitive to products’ quality in future study. Fourthly, in this study, the vendor is
the Stackelberg leader in the dual-channel supply chain. In future research, mod-
els with a leading retailer can also be studied.
Acknowledgements This work is supported by the National Natural Science Foundation of China (Nos.
71771053, 71628101, 71371003, and 71390334), Fundamental Research Funds for the Central Univer-
sities (No. 2242017K41036), Research and Innovation Program of Postgraduates in Jiangsu Province
(No. KYLX_0140), and Scientific Research Foundation of Graduate School of Southeast University
(No. YBJJ1526). The research has also been partly sponsored by Project 777742 (EC H2020-MSCA-
RISE-2017) and Project 691249 (EU H2020-MSCA-RISE-2015).
Appendix 1: Proof of Lemma 1

Proof The inventory level in the nth phase in time interval t ∈ [(n − 1)T, nT] satis-
fies the differential equation İ vn (t) = −𝜃Ivn (t) − Dv , t ∈ [(n − 1)T, nT] with boundary
condition Ivn (t = nT) = 0.
Solving the equation, the inventory level in the nth phase is
Dv 𝜃(nT−t)
Ivn (t) = (e − 1), t ∈ [(n − 1)T, nT]. (44)
𝜃
Then, solving the differential equation in the (n − 1)th phase, i.e.,

İ v(n−1) (t) = −𝜃Iv(n−1) (t) − Dv , t ∈ [(n − 2)T, (n − 1)T] with boundary condition
Iv(n−1) [t = (n − 1)T] − Ivn [t = (n − 1)T] = Qrn, the inventory level in time interval
t ∈ [(n − 2)T, (n − 1)T] can be solved as
Dv 𝜃(nT−t)
Iv(n−1) (t) = (e − 1) + Qrn e𝜃((n−1)T−t) , t ∈ [(n − 2)T, (n − 1)T]. (45)
𝜃
13
Following the same approach, the inventory level in time interval

t ∈ [(n − 3)T, (n − 2)T] is
Dv 𝜃(nT−t)
Iv(n−2) (t) = (e − 1) + Qrn e𝜃(n−1)T−t + Qr(n−1) e𝜃((n−2)T−t) , t ∈ [(n − 3)T, (n − 2)T].
𝜃
(46)
Finally, the inventory level in time interval t ∈ [0, T] is
Dv 𝜃(nT−t)
Iv1 (t) = (e − 1) + Qrn e𝜃((n−1)T−t) + Qr(n−1) e𝜃((n−2)T−t) + ⋯ + Qr2 e𝜃(T−t) , t ∈ [0, T].
𝜃
(47)
Based on the above analysis, the inventory level can be inducted as
Dv 𝜃(nT−t) dr
Ivj (t) = (e − 1) + (e(𝜃−𝜇)nT − e(𝜃−𝜇)jT )e−𝜃t , t ∈ [(j − 1)T, jT], j = 1, 2, … , n.
𝜃 𝜃−𝜇
(48)
This ends the proof of Lemma 1. □
Appendix 2: Proof of Proposition 1
Proof Before the proof, some definitions are made:

( )
1 − e−𝜇nT e𝜃nT − 𝜃nT − 1 1 1 − e−𝜇nT
X1 = , X2 = 2
, X3 = e(𝜃−𝜇)nT (1 − e−𝜃nT ) + (e𝜃T − 1) 𝜇T
,
𝜇nT 𝜃 nT (𝜃 − 𝜇)𝜃nT 1−e
( 𝜃T )
1 e − 1 e𝜇T − 1 1 − e−𝜇nT
X4 = − ,
(𝜃 − 𝜇)nT 𝜃 𝜇 e𝜇T − 1
1 𝜃nT 1
X5 = (e − 1), X6 = (e(𝜃−𝜇)nT − 1),
𝜃nT (𝜃 − 𝜇)nT
1 1 − e−𝜇nT
X7 = (e(𝜃−𝜇)T − 1) .
(𝜃 − 𝜇)nT 1 − e−𝜇T
The profit function can be rewritten as
Ar Av
TPsc = pv Dv + pr dr X1 − hv Dv X2 − hv dr X3 − hr dr X4 − cv Dv X5 − cv dr X6 − − .
T nT
(49)
(1) When n and T are fixed, taking the second order derivative of TPsc with respect
to pv and pr , the Hessian matrix can be obtained as
13
Y. He et al.
𝜕 2 TPsc 𝜕 2 TPsc ⎞ � �
−2b (1 + X1 )r
⎛
𝜕p2v 𝜕pv 𝜕pr
H=⎜ 𝜕 2 TPsc 𝜕 2 TPsc
⎟=
(1 + X1 )r −2b
. (50)
⎜ ⎟
⎝ 𝜕pr 𝜕pv 𝜕p2r ⎠
When the quality losing rate 𝜇 is not very large and b > r ,
𝜕 2 TPsc
|H| = 4b2 − (1 + X1 )2 r2 > 0 is satisfied. Also 𝜕p2v
= −2b < 0, so TPsc is jointly
concave in pv and pr . When equating both and to zero, the optimal prices
𝜕TPsc 𝜕TPsc
𝜕pv 𝜕pr
can be obtained by solving the equation set, which can be expressed as
B2 r + 2bB1 X1 + rX1 B2 − arX1 − arX12 + 𝛼arX12 − 2𝛼abX1 + 𝛼arX1

Pc∗
v =
,
r2 X12 − 4b2 X1 + 2r2 X1 + r2
(51)
2bB2 + rB1 + rB1 X1 − 2abX1 − 𝛼ar + 2𝛼abX1 − 𝛼arX1
Pc∗
r
= ,
r2 X12 − 4b2 X1 + 2r2 X1 + r2
(52)
in which B1 = −hv bX2 + hv rX3 + hr rX4 − cv bX5 + cv rX6,
B2 = hv rX2 − hv bX3 − hr bX4 + cv rX5 − cv bX6.
(2) For fixed pv, pr and n, there is only one decision parameter T.
Taking the second derivative of the profit function with respect to T, we have
𝜕 2 TPsc �� 2A 2A
2
= pr dr X1 − hv Dv X2 − hv dr X3 − hr dr X4 − cv Dv X5 − cv dr X6 − 3v − 3r .
𝜕T nT T
(53)
e−𝜇nT e−𝜇nT −1 xe−x +e−x −1
For X1 = . When setting x = −𝜇nT , X1 = . Defining a new
��
T
+ 𝜇nT 2 𝜇nT 2
function F(x) = xe + e − 1. When x → 0, X1 = 0. The first derivative of F(x)
��
−x −x
satisfies F (x) = −xe−x < 0. Thus in the interval T ∈ (0, +∞), X1 < 0 holds.
� ′′
For X2 = 𝜃2 nT1
(𝜃 2 n2 T 2 e𝜃nT − 2𝜃nTe𝜃nT + 2e𝜃nT − 2) , set x = 𝜃nT and define a
��
3
new function F(x) = x2 ex − 2xex + 2ex − 2 in which x ∈ (0, +∞). When x → 0,

F(x) = 0 and the first derivative satisfies F (x) = x2 ex > 0. Thus, in the interval
�
T ∈ (0, +∞), X2 > 0 always holds.

′′
The proofs of Xi > 0, (i = 3, 4, 5, 6) are the same as X2 > 0. Finally, the second
′′ ′′
𝜕 2 TPsc
derivative of profit function satisfies < 0, and TPsc is concave in T.
𝜕T 2
(3) Taking the second derivative of the profit function with respect to T, we have
𝜕 2 TPsc �� 2A
2
= pr dr X1 − hv Dv X2 − hv dr X3 − hr dr X4 − cv Dv X5 − cv dr X6 − 3 v .
𝜕n nT
(54)
13
e−𝜇nT e−𝜇nT −1 xe−x +e−x −1

For X1 = . When setting x = −𝜇nT , X1 = . Defining a new
��
n
+ 𝜇n2 T 𝜇nT 2
function F(x) = xe−x + e−x − 1. When x → 0, X1 = 0. The first derivative of F(x)

��
satisfies F (x) = −xe−x < 0. Thus in the interval T ∈ (0, +∞), X1 < 0 holds.
� ′′
For X2 = 𝜃2 n13 T (𝜃 2 n2 T 2 e𝜃nT − 2𝜃nTe𝜃nT + 2e𝜃nT − 2) , set x = 𝜃nT and define a

��
new function F(x) = x2 ex − 2xex + 2ex − 2, for x ∈ (0, +∞). When x → 0, F(x) = 0
and the first derivative satisfies F (x) = x2 ex > 0. Thus in the interval n ∈ (0, +∞)
�
,X2 > 0 holds.

′′
The proofs of Xi > 0, i = 3, 4, 5, 62 are the same as X2 > 0. Finally, the second
′′ ′′
derivative of profit function satisfies 𝜕n2 < 0, and TPsc is concave in n.

𝜕 TPsc
In the above analysis, we proved the existence and uniqueness for the decision
variables. To obtain the analytical results, we approximate the exponential terms and
solve equations 𝜕nsc = 0 and 𝜕nsc = 0, respectively. We use the Taylor expansion
𝜕TP 𝜕TP
to approximate exponential terms. For example, the term e𝜃T − 1 is approximated to

2 2 2 2
e𝜃T − 1 ≈ 1 + 𝜃T + 𝜃 2T − 1 = 𝜃T + 𝜃 2T . Thus, we obtain the final results in Propo-
sition 1.
This ends the proof of Proposition 1. □
Proof The vendor’s profit function can be expressed as

Av
TPv = pv Dv + wdr X7 − hv Dv X2 − hv dr X3 − cv Dv X5 − cv dr X6 − . (55)
nT
The retailer’s profit function can be expressed as
Ar
TPr = pr dr X1 − hr dr X4 − wdr X7 − . (56)
T
(1) When T and n are fixed, the value of Xi , (i = 1, 2, 3, 4, 5, 6, 7) are determined.

Equating the first derivative of retailer’s profit function to zero, the optimal price can
be derived as
1(
(57)
)
pr (pv , w) = (1 − 𝛼)a + rpv + bhr X4 + bX7 w .
2b
Substitute it into demand function, we have

( ) ( 2 )
r(1 − 𝛼) r 1 1
Dv (pv , w) = +𝛼 a+ − b pv + rhr X4 (T) + rwX7 (T), (58)
2b 2b 2 2
13
Y. He et al.
(1 − 𝛼)a 1 1 1
dr (pv , w) = + rpv − bhr X4 − bwX7 . (59)
2 2 2 2
Substitute the demand functions into vendor’s profit function, and take the first
𝜕 2 TP 𝜕 2 TP 2
derivative of TPv with respect to pv and w, there is 𝜕w2 v = −bX72, 𝜕p2 v = rb − 2b,
v
𝜕 2 TPv 𝜕 2 TPv
𝜕w𝜕pv
= 𝜕pv 𝜕w
= rX7. The Hessian matrix is
𝜕 2 TPv 𝜕 2 TPv � �
⎛
𝜕w2 𝜕w𝜕pv
⎞ −bX72 rX7
H=⎜ 𝜕 2 TPv 𝜕 2 TPv
⎟= r2 (60)
⎜ ⎟ rX7 b
− 2b
⎝ 𝜕pv 𝜕w 𝜕p2v ⎠
𝜕 2 TP
For 𝜕w2 v = −bX72 < 0, |H| = (b2 − r2 )X72 > 0, vendor’s profit function is jointly
concave in pv and w. And the optimal solution can be derived from the first order
conditions.
𝜕TPv (pv ,pr (pv ,w),w) 𝜕TP (p ,p (p ,w),w)
𝜕p
= 0, v v 𝜕wr v = 0. Then the optimal prices can be deter-
v
mined as
2bC2 + 2rC1 − 2𝛼ab − arX7 + 𝛼ar + 𝛼arX7
pd∗
v =
, (61)
2(r2 X7 − 2b2 + r2 )
2b2 C1 − r2 C1 − ab2 X7 + 𝛼ab2 X7 + brX7 C2 − 𝛼abrX7

wd∗ = , (62)
X7 b(r2 X7 − 2b2 + r2 )
1
pd∗
r
= ((1 − 𝛼)a + rpd∗
v
+ bhr X4 + bX7 wd∗ ), (63)
2b
2
in which ,
hr bX4 X7 h rX X h bX X cv rX5 X7 cv bX6 X7 r
C1 = 2
+ v 22 7 − v 23 7 + 2
− 2
C2 = hv X2 ( 2b − b)
hv rX3 r2 cX r h rX
+ +2
cv X5 ( 2b − b) + v 2 6 − r 2 4
(2)Taking the second derivative of retailer’s profit function with respect to T , we
have
𝜕 2 TPr 2A
(64)
��
= pr dr X1 − hr dr X4 − wdr X7 − 3r .
𝜕T 2 T
𝜕 2 TP
As proved in Proposition 1, X1 < 0 , X4 > 0 and X7 > 0 are satisfied. Thus 𝜕T 2 r < 0,
′′ ′′ ′′
and retailer’s profit function is concave in T.

(3)Taking the second derivative of vendor’s profit function with respect to n ,
13
𝜕 2 TPv 2A
(65)
��
= wdr X7 − hv Dv X2 − hv dr X3 − cv Dv X5 − cv dr X6 (n) − 3 r .
𝜕n2 nT
As proved in Proposition 1, Xi > 0.i = 2, 3, 5, 6. For

′′
−𝜇2 T 2 e−𝜇nT
1
. Finally, the second derivative of retailer’s
��
X7 = (𝜃−𝜇)T
(𝜃−𝜇)nT
(e − 1) 1−e−𝜇T
< 0
𝜕 2 TP
profit function satisfies 𝜕n2 v < 0 , thus TPv is concave in n. Although is an integer
variable, it is obvious that there exists a unique value of n that maximize the profit
function when pv, pr and T are constant.
In the above analysis, we proved the existence and uniqueness for the decision
variables. To obtain the analytical results, we approximate the exponential terms and
solve equations 𝜕T r = 0 and 𝜕n v = 0, respectively. We use the Taylor expansion to
𝜕TP 𝜕TP
approximate exponential terms. We obtain the final results in Proposition 2.

Proof For any given w, 𝛽 and F, the retailer’s retail price pr and ordering cycle T
should satisfy
𝜕TPco
r
|pr =pcr = 0, (66)
𝜕pr
𝜕TPco
r
|T=T c = 0. (67)
𝜕T
Taking the first order partial derivative of (31) with respect to pr and T and setting
them to zero yields
𝜕TPco
r
= (1 − 𝛽)(1 − 𝜇nT∕2)((1 − 𝛼)a − 2bpr + rpv ) + hr bT∕2 + wb(1 − (𝜃 − 𝜇)nT∕2) = 0,
𝜕pr
(68)
𝜕TPco
r
= (1 − 𝛽)pr dr (−𝜇n∕2) − hr dr ∕2 − wdr (−(𝜃 − 𝜇)n∕2) + Ar ∕T 2 = 0.
𝜕T
(69)
Substituting pr = and T = T into the two equations and solving (𝛽, w), the
pcr c
expressions of 𝛽 co and wco in Proposition 3 can be obtained. In addition, both par-

ties’ profit should be no less than that without coordination. Thus the lump sum fee
need to satisfy
13
Y. He et al.
F ∈ {TPco
r
r
, TPco
v
v
}. (70)
Proof Calculating the first and second order derivative of B(F) as follows
𝜕B(F)
= {(1 − 𝛾) TPco co co d∗
[ ] [ co co co d∗
]
𝜕F v (𝛽 , w , F = 0) − TPv − F − 𝛾 TPr (𝛽 , w , F = 0) − TPr + F }×
(71)
[ co co co ]𝛾−1 [ co co co ]−𝛾
TPv (𝛽 , w , F = 0) − TPd∗
v −F TPr (𝛽 , w , F = 0) − TPd∗
r .
𝜕 2 B(F) ]𝛾−2 [ co co co ]−𝛾−1

= −𝛾(1 − 𝛾) TPco co co d∗
TPr (𝛽 , w , F = 0) − TPd∗
[
v (𝛽 , w , F = 0) − TPv − F r +F ×
𝜕F 2
[ co co co 2
TPv (𝛽 , w , F = 0) − TPd∗ co co co d∗
]
v + TPr (𝛽 , w , F = 0) − TPr .
(72)
𝜕 2 B(F)
The second order derivative 𝜕F2 is negative, which means there exists an optimal F
that maximize the function. Equating 𝜕B(F)
𝜕F
to zero, the optimal F can be obtained as
F ∗ = TPco (𝛽 co , wco , F = 0) − TPd∗ ∗ (1 − 𝛾) − TPco (𝛽 co , wco , F = 0) − TPd∗

( ) ( )
r v r r
∗ 𝛾.
(73)
References
Baron O, Berman O, Wu D (2016) Bargaining within the supply chain and its implications in an industry.
Decis Sci 47:193–218
Batarfi R, Jaber MY, Zanoni S (2016) Dual-channel supply chain: a strategy to maximize profit. Appl
Math Model 40:9454–9473
Burwell TH, Dave DS, Fitzpatrick KE, Roy MR (1997) Economic lot size model for price-dependent
demand under quantity and freight discounts. Int J Prod Econ 48:141–155
Cai GG (2010) Channel selection and coordination in dual-channel supply chains. J Retail 86:22–36
Cai GG, Zhang ZG, Zhang M (2009) Game theoretical perspectives on dual-channel supply chain compe-
tition with price discounts and pricing schemes. Int J Prod Econ 117:80–96
Chauhan A, Singh AP (2015) Optimal replenishment and ordering policy for time dependent demand and
deterioration with discounted cash flow analysis. Int J Math Oper Res 6:407–436
Chen J, Liang L, Yao D-Q, Sun S (2017) Price and quality decisions in dual-channel supply chains. Eur J
Oper Res 269:935–948
Chen J, Zhang H, Sun Y (2012) Implementing coordination contracts in a manufacturer Stackelberg dual-
channel supply chain. Omega 40:571–583
Chen J-M, Chen T-H (2007) The profit-maximization model for a multi-item distribution channel. Transp
Res Part E: Logist Transp Rev 43:338–354
13
Chen TH (2015) Effects of the pricing and cooperative advertising policies in a two-echelon dual-channel
supply chain. Comput Ind Eng 87:250–259
Chiang WK, Chhajed D, Hess JD (2003) Direct marketing, indirect profits: a strategic analysis of dual-
channel supply-chain design. Manag Sci 49:1–20
Chiang WK, Monahan GE (2005) Managing inventories in a two-echelon dual-channel supply chain. Eur
J Oper Res 162:325–341
Choi T-M, Li Y, Xu L (2013) Channel leadership, performance and coordination in closed loop supply
chains. Int J Prod Econ 146:371–380
Dan B, Xu G, Liu C (2012) Pricing policies in a dual-channel supply chain with retail services. Int J Prod
Econ 139:312–320
Dumrongsiri A, Fan M, Jain A, Moinzadeh K (2008) A supply chain model with direct and retail chan-
nels. Eur J Oper Res 187:691–718
Dye C-Y, Chang H-J, Teng J-T (2006) A deteriorating inventory model with time-varying demand and
shortage-dependent partial backlogging. Eur J Oper Res 172:417–429
Dye C-Y, Hsieh T-P (2012) An optimal replenishment policy for deteriorating items with effective invest-
ment in preservation technology. Eur J Oper Res 218:106–112
Dye C-Y, Hsieh T-P (2013) A particle swarm optimization for solving lot-sizing problem with fluctuating
demand and preservation technology cost under trade credit. J Global Optim 55:655–679
Fibich G, Lowengart O, Gavious A (2003) Explicit solutions of optimization models and differential
games with nonsmooth (asymmetric) reference-price effects. Oper Res 51:721–734
Ghare P, Schrader G (1963) A model for exponentially decaying inventory. J Ind Eng 14:238–243
Giri BC, Jalan A, Chaudhuri K (2003) Economic order quantity model with Weibull deterioration distri-
bution, shortage and ramp-type demand. Int J Syst Sci 34:237–243
He R, Xiong Y, Lin Z (2016) Carbon emissions in a dual channel closed loop supply chain: the impact of
consumer free riding behavior. J Clean Prod 134:384–394
He Y, Zhang P, Yao Y (2014) Unidirectional transshipment policies in a dual-channel supply chain. Econ
Model 40:259–268
Hsu P, Wee H, Teng H (2010) Preservation technology investment for deteriorating inventory. Int J Prod
Econ 124:388–394
Hua G, Wang S, Cheng TE (2010) Price and lead time decisions in dual-channel supply chains. Eur J
Oper Res 205:113–126
Huang S, Yang C, Zhang X (2012) Pricing and production decisions in dual-channel supply chains with
demand disruptions. Comput Ind Eng 62:70–83
Huang W, Swaminathan JM (2009) Introduction of a second channel: implications for pricing and profits.
Eur J Oper Res 194:258–279
Ji J, Zhang Z, Yang L (2017) Carbon emission reduction decisions in the retail-/dual-channel supply
chain with consumers’ preference. J Clean Prod 141:852–867
Khuntonthong P, Chakpitak N, Neubert G (2013) Management framework for a high-value of agricultural
product to increase income for farmers in rural area. KMITL-Sci Technol J 13:42–50
Kopalle PK, Rao AG, Assuno JL (1996) Asymmetric reference price effects and dynamic pricing poli-
cies. Mark Sci 15:60–85
Lee JH, Moon IK (2006) Coordinated inventory models with compensation policy in a three level supply
chain. In: Computational science and its applications-ICCSA 2006. Springer, pp 600–609
Li B, Hou PW, Chen P, Li QH (2016a) Pricing strategy and coordination in a dual channel supply chain
with a risk-averse retailer. Int J Prod Econ 178:154–168
Li B, Zhu M, Jiang Y, Li Z (2016b) Pricing policies of a competitive dual-channel green supply chain. J
Clean Prod 112:2029–2042
Li Q-H, Li B (2016) Dual-channel supply chain equilibrium problems regarding retail services and fair-
ness concerns. Appl Math Model 40:7349–7367
Liang Y, Zhou F (2011) A two-warehouse inventory model for deteriorating items under conditionally
permissible delay in payment. Appl Math Model 35:2221–2231
Lin Y, Yu JC, Wang K-J (2009) An efficient replenishment model of deteriorating items for a supplier–
buyer partnership in hi-tech industry. Prod Plan Control 20:431–444
Lin Y-H, Lin C, Lin B (2010) On conflict and cooperation in a two-echelon inventory model for deterio-
rating items. Comput Ind Eng 59:703–711
Liu B, Zhang R, Xiao M (2010) Joint decision on production and pricing for online dual channel supply
chain system. Appl Math Model 34:4208–4218
13
Y. He et al.
Liu M, Cao E, Salifou CK (2015) Pricing strategies of a dual-channel supply chain with risk aversion.
Transp Res Part E: Logist Transp Rev 90:108–120
Lo S-T, Wee H-M, Huang W-C (2007) An integrated production-inventory model with imperfect produc-
tion processes and Weibull distribution deterioration under inflation. Int J Prod Econ 106:248–260
Lu Q, Liu N (2015) Effects of e-commerce channel entry in a two-echelon supply chain: a comparative
analysis of single- and dual-channel distribution systems. Int J Prod Econ 165:100–111
Mahata GC (2012) An EPQ-based inventory model for exponentially deteriorating items under retailer
partial trade credit policy in supply chain. Expert Syst Appl 39:3537–3550
Matsui K (2015) Asymmetric product distribution between symmetric manufacturers using dual-channel
supply chains. Eur J Oper Res 248:646–657
Matsui K (2017) When should a manufacturer set its direct price and wholesale price in dual-channel
supply chains? Eur J Oper Res 258:501–511
Moss CB, Schmitz TG, Kagan A, Schmitz A et al (2003) Institutional economics and the emergence of
E-commerce in agribusiness. J Agribus 21:83–102
Mukhopadhyay SK, Yao D-Q, Yue X (2008) Information sharing of value-adding retailer in a mixed
channel hi-tech supply chain. J Bus Res 61:950–958
Noh JS, Kim JS, Sarkar B (2016) Stochastic joint replenishment problem with quantity discounts and
minimum order constraints. Oper Res Int J. https://doi.org/10.1007/s12351–016–0281–6)
Qin Y, Wang J, Wei C (2014) Joint pricing and inventory control for fresh produce and foods with quality
and physical quantity deteriorating simultaneously. Int J Prod Econ 152:42–48
Rodriguez B, Aydin G (2015) Pricing and assortment decisions for a manufacturer selling through dual
channels. Eur J Oper Res 242:901–909
Sana S, Goyal S, Chaudhuri K (2004) A production-inventory model for a deteriorating item with trended
demand and shortages. Eur J Oper Res 157:357–371
Sarkar B (2011) An EOQ model with delay in payments and stock dependent demand in the presence of
imperfect production. Appl Math Comput 218:8295–8308
Sarkar B (2012a) An EOQ model with delay in payments and time varying deterioration rate. Math Com-
put Modell 55:367–377
Sarkar B (2012b) An inventory model with reliability in an imperfect production process. Appl Math
Comput 218:4881–4891
Sarkar B (2013) A production-inventory model with probabilistic deterioration in two-echelon supply
chain management. Appl Math Model 37:3138–3151
Sarkar B (2016) Supply chain coordination with variable backorder, inspections, and discount policy for
fixed lifetime products. Math Problems Eng 2016:1–14
Sarkar B, Majumder A, Sarkar M, Dey B, Roy G (2016) Two-echelon supply chain model with manufac-
turing quality improvement and setup cost reduction. J Ind Manag Optim 13:1085–1104
Sarkar B, Saren S, Cárdenas-Barrón LE (2015) An inventory model with trade-credit policy and variable
deterioration for fixed lifetime products. Ann Oper Res 229:677–702
Sarkar B, Saren S, Wee HM (2013) An inventory model with variable demand, component cost and sell-
ing price for deteriorating items. Econ Model 30:306–310
Sarkar M, Sarkar B (2013) An economic manufacturing quantity model with probabilistic deterioration in
a production system. Econ Model 31:245–252
Sarker BR, Mukherjee S, Balan CV (1997) An order-level lot size inventory model with inventory-level
dependent demand and deterioration. Int J Prod Econ 48:227–236
Shah NH, Soni HN, Patel KA (2013) Optimizing inventory and marketing policy for non-instantaneous
deteriorating items with generalized type deterioration and holding cost rates. Omega 41:421–430
Skouri K, Konstantaras I, Papachristos S, Ganas I (2009) Inventory models with ramp type demand rate,
partial backlogging and Weibull deterioration rate. Eur J Oper Res 192:79–92
Song HM, Zhao ZQ (2010) A study of Stackelberg inventory model with vendor-buyer interaction on
lead time. In: International conference on management and service science (MASS), 2010, IEEE,
pp 1–4
Sorger G (1988) Reference price formation and optimal marketing strategies. Optim Control Theory
Econ Anal 3:97–120
Takahashi K, Aoi T, Hirotani D, Morikawa K (2011) Inventory control in a two-echelon dual-channel
supply chain with setup of production and delivery. Int J Prod Econ 133:403–415
Thangam A, Uthayakumar R (2009) Two-echelon trade credit financing for perishable items in a supply
chain when demand depends on both selling price and credit period. Comput Ind Eng 57:773–786
13
Tsay AA, Agrawal N (2004) Channel conflict and coordination in the e-commerce age. Prod Oper Manag
13:93–110
Wang K-J, Lin Y, Jonas C (2011) Optimizing inventory policy for products with time-sensitive deteriorat-
ing rates in a multi-echelon supply chain. Int J Prod Econ 130:66–76
Wang X, Li D (2012) A dynamic product quality evaluation based pricing model for perishable food sup-
ply chains. Omega 40:906–917
Wilder C (1999) HP’s online push. InformationWeek 4:53–54
Xiao T, Choi T-M, Cheng T (2014) Product variety and channel structure strategy for a retailer-stackel-
berg supply chain. Eur J Oper Res 233:114–124
Xiao T, Shi J (2016) Pricing and supply priority in a dual-channel supply chain. Eur J Oper Res
254:813–823
Xie JP, Liang L, Liu LH, Ieromonachou P (2017) Coordination contracts of dual-channel with coopera-
tion advertising in closed-loop supply chains. Int J Prod Econ 183:528–538
Xu H, Liu ZZ, Zhang SH (2012) A strategic analysis of dual-channel supply chain design with price and
delivery lead time considerations. Int J Prod Econ 139:654–663
Yan B, Wang T, Liu Y-P, Liu Y (2016) Decision analysis of retailer-dominated dual-channel supply chain
considering cost misreporting. Int J Prod Econ 178:34–41
Yan R (2008) Profit sharing and firm performance in the manufacturer–retailer dual-channel supply
chain. Electron Commer Res 8:155–172
Yan R, Pei Z (2009) Retail services and firm profit in a dual-channel market. J Retail Consum Serv
16:306–314
Yang JQ, Zhang XM, Fu HY, Liu C (2017) Inventory competition in a dual-channel supply chain with
delivery lead time consideration. Appl Math Model 42:675–692
Yu D-Z, Cheong T, Sun D (2016) Impact of supply chain power and drop-shipping on a manufacturers
optimal distribution channel strategy. Eur J Oper Res 259:554–563
Yue X, Liu J (2006) Demand forecast sharing in a dual-channel supply chain. Eur J Oper Res
174:646–667
Zhang J, Liu G, Zhang Q (2015) Coordinating a supply chain for deteriorating items with a revenue shar-
ing and cooperative investment contract. Omega 56:37–49
13

Research Paper 3

Uploaded by

Copyright:

Available Formats

Research Paper 3

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Research Paper 3

Uploaded by

Copyright:

Available Formats

Oper Res Int J

Inventory and pricing decisions for a dual‑channel

Yong He1 · Hongfu Huang1 · Dong Li2

Received: 25 December 2016 / Revised: 19 December 2017 / Accepted: 11 March 2018

Abstract Dual-channel supply chain structure, i.e., a traditional retail channel

Keywords Dual channel supply chain · Product deterioration · Game theory ·

Table 1 Related literature on dual-channel supply chain

Chiang et al. (2003)

Chen et al. (2012)

Choi et al. (2013)

Cai et al. (2009)

Chiang and Monahan

Xie et al. (2017) Cooperation advertising

Yan and Pei (2009) Retailer service

Mukhopadhyay et al. Retailer service, information

Li and Li (2016) Retail service

Hua et al. (2010) Lead time

Xu et al. (2012) Lead time

Yang et al. (2017) Lead time, stochastic demand

Xiao et al. (2014) Product variety

Dumrongsiri et al. (2008) Stochastic demand

Xiao and Shi (2016) Uncertain supply, channel

Liu et al. (2015) Risk aversion

Yue and Liu (2006) Information sharing

Huang et al. (2012) Demand disruption

He et al. (2014) Transshipment, stochastic

Yu et al. (2016) Drop shipping strategy

Matsui (2015) Competition of two manu-

Lu and Liu (2015) External e-commerce com-

Li et al. (2016b) Product greenness

He et al. (2016) Product greenness, free-

Takahashi et al. (2011) Delivery decisions

Rodriguez and Aydin Assortment planning

This paper Product deterioration

Table 2 Related literature on deteriorating inventory models

Ghare and Schrader (1963) Quantity One

Lin et al. (2010) Quantity Two Cooperative decision

Hsu et al. (2010) Quantity One Preservation investment

Sarkar and Sarkar (2013) Quantity One Probabilistic deterioration

Chauhan and Singh (2015) Quantity One Cash flow discount

it can either be a constant parameter (Mahata 2012) or be a time dependent parame-

3 Model formulation and solution analysis

Fig. 1 Supply chain structure and inventory system

Notation in this paper are presented in Table 3.

Dr (pv , pr ) = (1 − 𝛼)a − bpr + rpv ]e−𝜇t , (2)

In this subsection, the retailer’s profit is calculated. According to previous studies on

The ordering quantities of the retailer in phase j can be obtained by equating t to

The retailer’s total inventory holding quantities in phase j is

(2) Retailer’s total holding cost ( HCr)

(3)Retailer’s total purchasing cost ( PCr)

(4) Retailer’s total ordering cost (OCr)

Lemma 1 The inventory level for the vendor in a cycle in phase j is

(3) Vendor’s total inventory holding cost ( HCv)

(4) Vendor’s total purchasing cost ( PCv)

(5) Vendor’s ordering cost (OCv)

3.5 Analysis of the centralized problem

In this subsection, a centralized dual-channel supply chain is considered, in which

TPsc (pv , pr , n, T) = TPv + TPr

Proposition 1 When 𝜃 and 𝜇 are relatively small,

B2 r + 2bB1 X1 + rX1 B2 − arX1 − arX12 + 𝛼arX12 − 2𝛼abX1 + 𝛼arX1

3 Model formulation and solution analysis

3.5 Analysis of the centralized problem

3.6 Analysis of the decentralized problem

4.1 Examples for centralized and decentralized models

4.2 Sensitivity analysis on equilibrium strategies

5 Supply chain coordination

5.2 The optimal lump sum fee

6 Extension: delivery time as a decision variable