Anda di halaman 1dari 23

Ann Oper Res (2016) 241:475–496

DOI 10.1007/s10479-012-1085-6

A two-stage supply chain with demand sensitive to price,


delivery time, and reliability of delivery

Tiaojun Xiao · Xiangtong Qi

Published online: 11 February 2012


© Springer Science+Business Media, LLC 2012

Abstract We consider a two-stage supply chain with one supplier and one manufacturer.
The manufacturer faces a Poisson demand process where the arrival rate depends on the
selling price, the announced delivery time, and the delivery reliability defined as the proba-
bility of satisfying the announced delivery time. Such a demand model generalizes the works
in the literature by simultaneously considering the above three demand sensitivity factors.
The main purpose of this paper is to study the equilibrium decisions in the supply chain
with an all-unit quantity discount contract. We consider four scenarios regarding whether
the leadtime standard, the delivery reliability standard, and the manufacturer’s capacity are
endogenous, and whether the manufacturer’s production cost is its private information. We
find that an all-unit quantity discount scheme can coordinate the supply chain for most cases.
Managerial insights are observed regarding the impact of the three demand sensitivity fac-
tors. For example, the breakpoint in an optimal quantity discount contract always increases
with the delivery reliability sensitivity under an exogenous delivery reliability, but may de-
crease under an endogenous delivery reliability; with asymmetric information, a higher vari-
ance of the manufacturer’s unit production costs leads to a lower unit wholesale price for
the low-cost manufacturer.

Keywords Supply chain coordination · Price quotation · Leadtime quotation · Delivery


reliability

T. Xiao
School of Management Science and Engineering, Nanjing University, Hankou Road 22, Nanjing,
Jiangsu 210093, China
e-mail: xiaotj@nju.edu.cn

B
X. Qi ( )
Department of Industrial Engineering and Logistics Management, The Hong Kong University
of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China
e-mail: ieemqi@ust.hk
476 Ann Oper Res (2016) 241:475–496

1 Introduction

Demand shaping is one of the challenging problems in operations management. It has been
well recognized that the customers’ demand is possibly affected by a number of factors
such as the selling price, the delivery leadtime, and others. While the selling price has a
well-understood impact on the customer’s purchase decision, the leadtime factor may be
more complicated, especially in a make-to-order environment with random customer ar-
rivals and order processing times. A common business practice in many manufacturing and
service industries is that an expected leadtime is announced by the manufacturer or service
provider. For example, Dell will give a customer an estimated shipping date after the cus-
tomer configures a computer at www.dell.com, and a restaurant will offer a delivery time to
a customer who calls in to place an order for delivery. Intuitively, a short announced delivery
time would have a positive impact to attract more customer demand. However, the question
is whether this is always the case.
As pointed out by Shang and Liu (2011), an announced leadtime may not be a 100%
service guarantee. Due to various uncertainties from the demand arrival, production, and
delivery processes, some customers may receive their orders later than the announced de-
livery time, especially when the announced deliver time is too tight. A broad range of em-
pirical data presented in Shang and Liu (2011) (and the references therein) shows that late
deliveries frequently exist in many different industries. Late deliveries lead to a deteriorated
delivery reliability to customers, and will have a long-term negative effect on customers de-
mand. Therefore, the reliability of meeting the delivery time needs to be considered when a
company makes the decision of delivery time announcement.
In this paper, we will consider a two-stage supplier-manufacturer supply chain where
the downstream manufacturer’s demand is sensitive to three factors, the selling price of the
product, the announced delivery leadtime, and the delivery reliability standard that is defined
as the minimum probability of meeting the announced leadtime. As far as we known, this is
the first work that addresses these three factors simultaneously, though there exist researches
on joint decisions on pricing and leadtime, or joint decisions on announced leadtime and
delivery reliability standard. The need of such a more comprehensive framework to model
demand sensitivities in multiple dimensions is obvious, as commented in Shang and Liu
(2011):
It is worth pointing out that an ideal analytical model for such time-based competition
should incorporate pricing decisions. However, allowing for pricing will make the
model intractable.
Our problem, under the context of supply chain coordination, is more challenging be-
cause it further involves two additional factors, the upstream supplier’s wholesale pricing
policy and the downstream manufacturer’s processing capacity (service rate).
We will address the equilibrium decisions and channel coordination for the above
supplier-manufacturer system. In the traditional supplier-retailer channel, it is well known
that a fixed unit wholesale price makes the retailer’s pricing decision different from the
optimal retail price that maximizes the total channel profit, generally known as the double
marginalization phenomenon. To resolve such a problem, the supplier can design some coor-
dination schemes, such as a quantity discount wholesale price policy or a buyback policy, to
induce the retailer to make the channel optimal pricing decision. In a supplier-manufacturer
channel with demand sensitive to multiple factors, the coordination becomes more compli-
cated, i.e., the supplier must design some coordination schemes to induce the manufacturer
to make the channel optimal pricing, delivery leadtime and delivery reliability decisions.
Ann Oper Res (2016) 241:475–496 477

One of our major purposes is to derive equilibrium decisions and show how the supplier
can offer a specific quantity discount wholesale contract, one of the most commonly used
coordinating contracts, to coordinate the channel under different scenarios.
In the baseline model, we assume that the manufacturer’s delivery reliability standard and
unit production cost are known to the supplier. For example, there may be a pre-specified
company standard regulating the delivery reliability, which is a prevailing industrial stan-
dard. We first design a quantity discount wholesale contract to coordinate the channel, then
investigate how the channel performance and equilibrium outcome are affected by the chan-
nel parameters.
To extend the baseline model, we consider four alternative scenarios, a model with asym-
metric information on the manufacturer’s production cost, a model with an exogenous lead-
time standard, a model with an endogenous delivery reliability standard, and a model with
an endogenous capacity. We show that the supply chain can still be coordinated for all these
four cases, and discuss how the equilibrium outcome sensitivity is similar or different com-
pared with the baseline model.
The remainder of the paper is organized as follows. The related literature is reviewed
in Sect. 2. We present the baseline model and assumptions in Sect. 3. We then study the
coordination mechanism for the baseline model in Sect. 4, and four alternative models in
Sect. 5. We conclude the paper in Sect. 6.

2 Literature review

This paper is closely related to supply chain coordination management, pricing and service
level (leadtime and delivery reliability standard), capacity decisions, and asymmetric infor-
mation.

2.1 Coordination mechanism

There are extensive works on supply chain coordination under different settings. Various
contracts have been developed in the supply chain coordination literature. For a compre-
hensive introduction and review, we refer to Cachon (2004). Most of the existing works,
however, discuss the coordination for a supplier and a retailer, leaving the case for a supplier
and a Make-To-Order (MTO) manufacturer little addressed. The purpose of this paper is to
fill the void.
An all-unit quantity discount coordination mechanism has been discussed for supply
chain coordination in different environments. For example, Weng (1995) considered channel
coordination in a system consisting of a supplier and a group of homogeneous buyers, and
Chen et al. (2001) extended Weng’s model to non-identical retailers. Kolay et al. (2004)
found that the all-unit quantity discount can resolve the problem of double marginalization
when the demand only depends on the retail price and the demand information is symmetric.
Differing from the above works, we consider how to coordinate a channel consisting of
one supplier and one manufacturer who jointly determines the resale price and lead-time,
by employing the all-unit quantity discount scheme. Under our discount scheme, the new
challenge in coordination is that we need to consider the manufacturer’s decision in multiple
dimensions, such as leadtime quotation, price quotation, delivery reliability standard, and
capacity setting, which is more complicated than the coordination with a traditional retailer.
478 Ann Oper Res (2016) 241:475–496

2.2 Price, service level (leadtime/delivery reliability standard) and capacity decisions

Besides price and quality, delivery leadtime is another important factor that influences a
firm’s performance or consumers’ demands (Upasani and Uzsoy 2008). There exist two
streams of research in the literature regarding the delivery leadtime decision and the im-
pact on the consumer demand. The first stream assumes that demand is independent of the
leadtime (e.g., Yano 1987; Song et al. 2000; Elhafsi 2002; Ha et al. 2003), and the second
stream, which includes our work, assumes that demand is sensitive to the leadtime deci-
sion (e.g., Palaka et al. 1998; Rao et al. 2005; Liu et al. 2007; Yang and Geunes 2007;
Allon and Federgruen 2007; Pekgün et al. 2008). The closest work to our paper is the de-
termination of both pricing and leadtime decisions in a decentralized way. In this category,
Liu et al. (2007) considered a supply chain where the supplier determines the leadtime and
the retailer determines the retail price. Pekgün et al. (2008) assumed that, in a company, the
production department determines leadtime and the marketing department determines the
retail price, and studied how to coordinate both departments. Xiao et al. (2011) considered
a supply chain where the manufacturer determines the unit wholesale price and leadtime
standard and the retailer determines the retail price. Differing from the above models, we
consider a new case where the downstream manufacturer jointly determines the resale price
and leadtime decisions and investigate how the upstream supplier coordinates the supply
chain. As far as we known, such a setting has not been studied in the literature.
In the literature, service level can be defined in different ways. In a Make-To-Stock
(MTS) environment, service level is often described by fill rate, i.e., the probability of ful-
filling an order from on-hand inventory (Dana 2001; Benjaafar et al. 2007); In the MTO
environment, service level is often described by the probability of fulfilling requests within
a fixed quoted lead-time (Benjaafar et al. 2007); and in a service industry, service level
is often described by waiting time (Gilbert and Weng 1998; Allon and Federgruen 2007).
In this paper, we measure service level by two dimensions, delivery leadtime and delivery
reliability, which makes the coordination more challenging.

2.3 Asymmetric information

In the supply chain management literature, asymmetric information often means that the
downstream firm(s) has certain private information on its cost structure (Corbett and de
Groote 2000; Corbett 2001; Ha 2001; Corbett et al. 2004) or the market demand (Desiraju
and Moorthy 1997; Özer and Wei 2006). Owing to space limit, we can only list a few works
here. For example, Corbett (2001) developed principal-agent models to study the effects
of information asymmetries about setup cost and backorder cost, respectively. Desiraju and
Moorthy (1997) investigated how performance (price or service) requirements may improve
the working of a distribution channel. Kolay et al. (2004) designed a revelation mechanism
of all-units discounts under asymmetric demand information from the manufacturer’s per-
spective. The revelation mechanism may not maximize the channel profit due to the absence
of coordination. Unlike Kolay et al. (2004), we focus on how to coordinate the supply chain
under the asymmetric information on the manufacturer’s unit production cost from the third
party’s perspective and the effect of the cost variance on the coordination mechanism.

3 The basic model

Consider a supply chain consisting of one supplier and one manufacturer. The supplier pro-
duces a standard product in the MTS mode. The manufacturer purchases standard products
Ann Oper Res (2016) 241:475–496 479

from the supplier at a unit wholesale price, takes orders from end users, customizes the
standard products based on order specifications, and delivers the final products to end users.
The supplier and the manufacturer maximize their long-term average profit per unit time,
respectively. Under demand uncertainty, vendor managed inventory (VMI) provides posi-
tive benefits to supply chain participants by reducing inventory cost (Zhang et al. 2007). We
assume that the supply chain adopts VMI mode to manage the supply chain’s inventory, and
the production rate of the supplier is greater than the manufacturer’s service rate (capacity).
The manufacturer operates in the make-to-order (MTO) mode. Specifically, consumer or-
ders arrive randomly and it takes the manufacturer a random time to process each consumer
order. Under VMI, the effect of resource constraint on the manufacturer can be ignored be-
cause the MTS supplier has a greater production rate than the manufacturer’s capacity and
can well control the inventory of standard products. Hence, similar to Gupta and Benjaafar
(2004) and Palaka et al. (1998), we can approximately use the M/M/1 queuing model to
describe the operations of the manufacturer where orders arrive as a Poisson process with
arrival rate λ and the processing time of each order follows an exponential distribution with
service rate μ. Such an M/M/1 queuing model is a reasonably simple yet acceptable for-
mulation. In the basic model, we assume that the manufacturer’s capacity has been fixed
before designing the coordination mechanism, which is a common assumption in the liter-
ature because building capacity needs a long lead-time (Liu et al. 2007). Specifically, the
supplier first determines the unit wholesale price, and then the manufacturer determines the
resale (retail) price and leadtime standard. We will extend it to the case where the manufac-
turer determines the resale price, leadtime standard, and delivery reliability standard in an
alternative model. We also discuss the case where the manufacturer’s capacity is determined
after the coordination mechanism was designed in an alternative model with an endogenous
capacity.
We summarize below the notation used in our formulation.
a0 : basic market scale (basic demand curve intercept), a0 > 0
cS : unit production cost of the supplier, including inventory holding cost
cM : unit production cost of the manufacturer, excluding the unit wholesale price charged
by the supplier
cμ : unit capacity cost of the manufacturer
w: unit wholesale price charged by the supplier, w ≥ cS
p: resale price quoted by the manufacturer, p ≥ cM + w
L: delivery leadtime standard quoted by the manufacturer, L > 0; 1/L describes the man-
ufacturer’s service level
α: price sensitivity of demand, α > 0
β: leadtime sensitivity of demand, β > 0
γ : delivery reliability sensitivity of demand, γ > 0
λ: demand rate of the consumers, λ > 0
μ: service rate (capacity) of the manufacturer, μ > 0
s: delivery reliability standard of the manufacturer under delivery time guarantee, 0 < s <
1
k: used for computational simplicity where k = ln(1/(1 − s)) is increasing with s
t : the fraction of the manufacturer’s profit in the channel profit, 0 ≤ t ≤ 1
The demand, i.e., order arrival rate λ, is sensitive to both the resale price quotation p
and the leadtime quotation L. Due to the queuing randomness, a quoted leadtime L cannot
be always guaranteed. For any quoted positive leadtime L, we can calculate a probability
(reliability) of the real delivery time being no later than L for any order. From the M/M/1
480 Ann Oper Res (2016) 241:475–496

queuing model, we know that the realized delivery reliability for given lead-time L is 1 −
e−(μ−λ)L . Thus, given the delivery reliability standard s, the manufacturer should quote the
lead-time L satisfying the constraint 1 − e−(μ−λ)L ≥ s, i.e., the probability of meeting the
quoted lead-time should be at least as large as the required delivery reliability standard
(Palaka et al. 1998; Pekgün et al. 2008). For notational convenience, we define k = ln(1/
(1 − s)) where clearly k is increasing with s and k > 0. Furthermore, we can rewrite the
constraint 1 − e−(μ−λ)L ≥ s as
(μ − λ)L ≥ k. (1)
This constraint is referred to as the delivery reliability constraint. It describes the relationship
among the demand rate, the lead-time and the delivery reliability standard.
In general, the delivery reliability standard also affects the demand rate because frequent
late deliveries definitely have an adverse impact on the consumers purchasing decision.
A higher delivery reliability standard s increases the market share (Shang and Liu 2011).
Therefore, we assume that it has a positive effect on demand rate such as
λ = a0 + γ · s − α · p − β · L. (2)
In (2), the demand rate λ is decreasing with the price quotation p and the leadtime quota-
tion L, and increasing with the delivery reliability standard s. The linear relationship among
λ, p, and L is a common assumption in the literature; for example, see Liu et al. (2007). We
further extend it to the case linearly depending on the delivery reliability standard s. In this
way, we can regard a0 as the basic market scale of consumers.
In the major part of this paper, we assume that the delivery reliability standard s is prede-
termined by the marketing and operations decisions. This may happen, for example, when
the manufacturer regards the probability of achieving the promised delivery leadtime stan-
dard as a company target that represents the long-term reputation. In such a case, a0 + γ · s
becomes a constant that can be denoted by a(s) = a0 + γ · s. We refer to a(s) as the market
scale (demand curve intercept) under the given delivery reliability standard s. Now, we can
rewrite (2) into
λ = a(s) − α · p − β · L. (3)
We will consider the case where the delivery reliability standard s is an endogenous variable
in an alternative model (see Sect. 5).
The above modeling approach is similar to the related research involving price- and
leadtime-sensitive demand, e.g., in Liu et al. (2007) and Pekgün et al. (2008). Our model is
new in the way of handling the impact of late delivery. In Liu et al. (2007), each consumer
is compensated with a late penalty if the delivery is later than the quoted leadtime. As a
result, late delivery has no direct impact on the demand rate. Different from them, we do not
assume any penalty of late delivery; but frequent late delivery causes a goodwill loss which
will be eventually reflected by the reduction of the demand rate in the long run, as modeled
by the γ · s term in our model. The delivery reliability constraint (1) comes from Pekgün et
al. (2008), and we make it more general by including the possible negative impact of the late
delivery through the γ · s term. In fact, our model includes Pekgün et al. (2008) as a special
case with s being a constant.
To avoid trivial cases, we assume that the basic market scale a0 is sufficiently large.
Specifically, there is a positive demand rate for the manufacturer when the supplier offers
the unit wholesale price at the unit production cost w = cS , and the manufacturer quotes the
lowest resale price at p = cM + cS and the shortest leadtime L = k/μ, which implies that
a0 > α · (cM + cS ) + β · k/μ − γ · s.
Ann Oper Res (2016) 241:475–496 481

4 Coordination of the supply chain

4.1 Reaction of the manufacturer to the unit wholesale price

In the basic model, we consider the case where the manufacturer has a predetermined de-
livery reliability standard s. To study coordination, we first consider the optimal reaction of
the manufacturer to a given unit wholesale price.
With the given unit wholesale price w, the manufacturer determines the resale price p
and the quotation of the leadtime L to maximize the long-term average profit per unit time,
denoted by πM (L, p; w) = λ · (p − cM − w), by solving the following optimization problem.
max πM (L, p; w)
L,p
s.t. (μ − λ)L ≥ k. (4)
λ = a(s) − α · p − β · L.
Note that the optimization problem (4) is similar to the centralized decision model in
Pekgün et al. (2008). Proposition 1 summarizes the optimal solution of (4).

Proposition 1 The optimal reactions of the manufacturer to the unit wholesale price w
are p(w) = a(s)−λ(w)
α
− α·(μ−λ(w))
k·β
and L(w) = μ−λ(w)
k
, where λ(w)(∈ (0, μ)) is the optimal
demand rate that can be uniquely solved from the following first-order condition
    2
f1 w, λ(w) = a(s) − α · (cM + w) − 2λ(w) μ − λ(w) − k · β · μ = 0. (5)

Proposition 1 implies that the manufacturer’s resale price and leadtime decisions are
uniquely determined by w, the unit wholesale price charged by the supplier. This gives
the supplier an opportunity to manipulate the wholesale price to coordinate the channel. In
the following, we will use πM (L(w), p(w); w) to denote the maximum manufacturer profit
under a given w, with the dependency of L(w) and p(w) on w. The following result shows
how they will change with w.

Corollary 1 The manufacturer’s optimal leadtime quotation L(w) is decreasing with w,


price quotation p(w) is increasing with w, and the resulting demand rate λ(w) is decreasing
with w.

Corollary 1 (proof see in Appendix, etc.) implies that, when the unit wholesale price w
increases, the manufacturer will correspondingly raise the resale price, which results in a
lower demand rate. This is consistent with our intuition. At the same time, the manufac-
turer will quote a shorter leadtime, trying to offset the impact of the higher price on the
demand reduction. In this sense, we can say that the roles of price and leadtime decisions
are complementary with each other such that the overall effect can be balanced.
Given the delivery reliability standard s predetermined, we now discuss the impact of
changing s. Roughly speaking, the increase of the delivery reliability standard s has two
opposite effects on the demand rate. On one hand, when s increases, the manufacturer has
the option to quote a longer delivery leadtime in order to have the higher probability of
meeting the leadtime quotation, which makes the demand rate lower. On the other hand,
higher delivery reliability standard will raise the market scale, thanks to the better reputation.
As a result, when s increases, whether the demand rate is to increase or decrease will depend
on which of the above two factors is stronger.
482 Ann Oper Res (2016) 241:475–496

Corollary 2 Under a given fixed unit wholesale price w, the demand rate λ(w) is an in-

creasing function of s if λ(w) ≤ μ − βμ/[γ (1 − s)] and a decreasing function of s if

λ(w) > μ − βμ/[γ (1 − s)].


Corollary 2 can be further explained as follows. The threshold μ − βμ/[γ (1 − s)]
defines a downward boundary for the demand rate λ(w). If, at s = 0, λ(w) is above the
boundary, then λ(w) will be decreasing with s, but never going below the boundary. In
such a case, increasing the delivery reliability standard can only reduce the demand rate.
If, at s = 0, λ(w) is below the boundary, then λ(w) will first be increasing with s until it
crosses the boundary and starts to decrease. In other words, increasing the delivery reliability
standard will first increase the demand rate, and then decrease the demand rate.

4.2 Channel coordination

Note that the entire channel profit is given by πC (L, p) = λ · (p − cS − cM ). When the unit
wholesale price equals the unit production cost (w = cS ), the decentralized channel is equiv-
alent to the centralized channel. Therefore, for the channel, according to Proposition 1, we
have the optimal resale price at p(cS ), the optimal leadtime at L(cS ), the resulting demand
rate at λ(cS ), and the maximum channel profit at πM (L(cS ), p(cS ); cS ). We denote the opti-
mal demand rate for the centralized channel by λ1 = λ(cS ), and the maximum channel profit
by πC∗ = πM (L(cS ), p(cS ); cS ).
Given a fixed unit wholesale price w, unless w = cS , the manufacturer will make the
resale price and leadtime decisions away from the system optimum, under which the max-
imum channel profit cannot be achieved. The problem of channel coordination is how to
design a mechanism that induces the manufacturer to make the optimal decisions for the
entire channel. Specifically, the coordination mechanism should induce the manufacturer to
offer the resale price p(cS ) and the leadtime L(cS ). Here, mechanism designer adjusts the
fraction t to allocate the channel profit between the supplier and the manufacturer.
It is well known that an all-unit quantity discount scheme can be used to coordinate
the traditional supplier-retailer channel where the traditional retailer is induced to make the
optimal retail price for the channel (e.g., Kolay et al. 2004). A standard all-unit quantity
discount scheme includes three parameters λ0 , w1 and w2 (w1 < w2 ), in which the unit
wholesale price is w1 if the order volume achieves at least λ0 , otherwise, the unit wholesale
price is w2 . In this paper, the quantity discount applies to a unit time period, i.e., the discount
is determined by the total quantity purchased over the unit time period. Under VMI, the
total quantity of the manufacturer is equal to the demand rate. Thus, we can substitute the
demand rate for the order quantity in the all-unit quantity discount scheme. Now the question
is whether an all-unit quantity discount scheme can coordinate the supplier-manufacturer
channel in which the manufacturer has to jointly make two decisions, the resale price and
the leadtime quotation.
Recall that, when the channel is coordinated, the maximum channel profit is πC∗ . For
an ideal coordination scheme, we hope that the scheme can allocate the channel profit be-
tween the two players in any arbitrary ratio. Suppose that the manufacturer is to get t · πC∗ ,
0 < t < 1, in the coordinated channel. Then the mechanism designer can offer a quantity
discount contract defined in Proposition 2.
Ann Oper Res (2016) 241:475–496 483

Proposition 2 With complete information, there is an all-unit quantity discount mechanism


that coordinates the supplier-manufacturer channel in which the unit wholesale price w∗
depends on the order quantity (realized demand rate λ) such that
 ∗
∗ w1 , if λ ≥ λ1
w =
w2∗ , otherwise,
where w1∗ = (a(s) − λ1 )(1 − t)/α + cS · t − cM · (1 − t) − k · β · (1 − t)/[α · (μ − λ1 )] =
cS · t + (1 − t)(p(cS ) − cM ), w2∗ is any value greater than ŵ2 , and ŵ2 is the unique solution
that satisfies πM (L(ŵ2 ), p(ŵ2 ); ŵ2 ) = t · πC∗ .

In Proposition 2, as long as w2∗ is greater than ŵ2 , the manufacturer has no incentive
to choose an order quantity lower than λ1 . In other words, the price w2∗ is not unique and
ŵ2 is a lower bound of w2∗ . Note that the proof of Proposition 2 indicates that ŵ2 > w1∗ ,
i.e., the mechanism is a valid quantity discount scheme. The all-unit quantity discount
in Proposition 2 can also be presented by a simpler form such that w1∗ = η · w2∗ with
0 < η < 1, as shown in Kolay et al. (2004). Proposition 2 gives the closed form expression
of w1∗ as a function of the problem parameters, which allows us to investigate the effects of
the interaction between the resale price and lead-time and other factors on the coordination
mechanism. In the following, we will focus on such effects to derive some new managerial
insights.

Corollary 3 Under the all-unit quantity discount mechanism w∗ , the equilibrium profit of
each player is a decreasing function of the price sensitivity α, the leadtime sensitivity β,
and the unit production costs cM and cS , and an increasing function of the market scale
a0 and the manufacturer’s capacity μ. Moreover, the equilibrium profit of each player is an
increasing function of the delivery reliability standard s if and only if s < 1 − β/[γ (μ − λ1 )].

Most of the claims in Corollary 3 are quite intuitive. They suggest some potential ways
to increase the channel profit, for example, by cost reduction, marketing promotion, and
manufacturer capacity expansion. A more interesting result is the impact of the delivery
reliability standard on the channel profitability. We find that a higher delivery reliability
standard s leads to a higher channel profit only when the delivery reliability standard is
sufficiently low, quantified by s < 1 − β/[γ (μ − λ1 )]. This can be explained by the dual
effects of the delivery reliability standard on the demand rate. On one hand, higher delivery
reliability standard brings a larger market scale, a positive factor potentially increasing the
demand rate; on the other hand, a higher delivery reliability standard enforces a stronger
constraint on pricing and leadtime quotations, a negative factor potentially reducing the
demand rate. Moreover, the negative effect increases with the delivery reliability standard.
Therefore, only the increase of a low delivery reliability standard is beneficial to the channel.

4.3 Sensitivity analysis

To better understand the equilibrium outcome of the coordinated channel, we carry out a
sensitivity analysis by numerical experiments. In Table 1, we show how the equilibrium
outcome changes with different parameters, where the default values of parameters are
a0 = 40, γ = 8, cM = cS = 5, s = 0.95, t = 0.6, α = 1, β =1 and μ = 50.
For each parameter, we change it from its default value in a certain range, then calculate
the new channel equilibrium, and observe how the outcome may change. To simplify the
484 Ann Oper Res (2016) 241:475–496

Table 1 Sensitivity analysis on


the equilibrium outcome of the λ1 w1∗ p∗ L∗ πC∗
coordinated supply chain
a0 ↑ ↑ ↑ ↑ ↑ ↑
cM ↑ ↓ ↓ ↑ ↓ ↓
cS ↑ ↓ ↑ ↑ ↓ ↓
s↑ ↑ ↑ ↑ ↑ ↑

α↑ ↓ ↓ ↓ ↓ ↓
β↑ ↓ ↓ ↓ ↓ ↓
μ↑ ↑ ↓↑ ↓↑ ↓ ↑
Note: ↑ represents increase and ↓ γ↑ ↑ ↑ ↑ ↑ ↑
represents decrease

presentation, we only report whether a particular outcome is increasing (denoted by ↑) or


decreasing (denoted by ↓) when a parameter increases. The changing ranges of the param-
eters are
a0 ∈ [21, 45], γ ∈ [2, 50], cM , cS ∈ [1, 10], s ∈ [0.5, 0.995],
α, β ∈ [0.1, 2.5] and μ ∈ [20, 100].
We have also tested other values in our experiments, but there is no change on the major
observations.
We can highlight the following observations from Table 1.
(i) The equilibrium demand rate λ1 and channel profit πC∗ always change in the same
direction. In other words, if the equilibrium demand rate is increased due to any reason,
it must bring in more channel profit.
(ii) The supplier’s wholesale price w1∗ and the manufacturer’s resale price p ∗ change in the
same direction for most cases except the manufacturer cost change. It is quite intuitive
that the manufacturer adjusts p ∗ according to the change of w1∗ . However, for the case
of manufacturer cost increases, the manufacturer will respond by increasing the resale
price, which leads to demand rate decrease. To balance off such a change, the supplier
decreases the wholesale price to share the increased manufacturer’s cost so that the
manufacturer will not increase p ∗ too much.
(iii) When the channel cost increases, the leadtime standard L∗ and resale price p ∗ change
complementarily, i.e., in different directions. When the market demand rate related
parameters a0 , s, γ , α, and β change, they change in the same direction.
(iv) The prices first decrease and then increase with the manufacturer’s capacity. Specifi-
cally, when a low manufacturer’s capacity increases, raising demand rate has a large
benefit such that the supply chain decreases the retail price and lead-time standard to
stimulate demand rate. However, when a high manufacturer’s capacity increases, rais-
ing demand rate has a small benefit such that the supply chain reduces the lead-time
standard to stimulate demand rate but raising the prices to achieve a higher unit profit.

5 Robustness of coordination mechanism

In this section, we extend the basic model to four alternative cases, with the asymmetric
information, with an exogenous leadtime standard, with an endogenous delivery reliability
standard, and with an endogenous capacity, denoted by a suffix “Ai ”, i = 1, 2, 3, 4, respec-
tively.
Ann Oper Res (2016) 241:475–496 485

5.1 Asymmetry of information and channel coordination

In Sect. 4, we assume that the supplier holds the complete information of the manufacturer.
In practice, the manufacturer may conceal some information for a higher bargaining power.
In particular, the manufacturer’s unit production cost is likely to be the private information
of the manufacturer because it is not directly observable to the supplier. If the manufacturer
reveals wrong information to the supplier, the profit of the supplier and the channel will
be deteriorated. It is well known that, under asymmetric information, an expected profit-
maximizing revelation mechanism is more beneficial to the supplier than a coordination
mechanism because the latter is a special case of the former. Hence economists often focus
on revelation mechanisms rather than coordination mechanisms, e.g., Kolay et al. (2004).
However, a revelation mechanism may not maximize the channel benefit of a supply chain
due to the absence of coordination. Moreover, in reality, the supplier often shares the benefit
of coordination with the manufacturer. Thus, in this paper, we develop a coordination mech-
anism from the third party’s perspective: a supply chain manager or a supplier-manufacturer
coalition, e.g., headquarters of the decentralized firm where the supplier is production de-
partment and the manufacturer is marketing department (e.g., Pekgün et al. 2008). We will
discuss how the information asymmetry changes the coordination mechanism design.
To model information asymmetry, we assume that the manufacturer can be in one of
the two types with respect to the unit production cost cM , a low cost cM , and a high cost
c̄M , 0 < cM < c̄M . This way of describing information asymmetry is common in the eco-
nomics literature (Desiraju and Moorthy 1997). Similar to Corollary 1, we can show that
λA1 (wA1 ) is a decreasing function of cM . Thus, the type-cM manufacturer has a higher de-
mand rate reaction than the type-c̄M manufacturer. We rewrite their demand rate reactions
as λA1 (wA1 ) and λ̄A1 (wA1 ), respectively. Thus, we have λA1 (wA1 ) > λ̄A1 (wA1 ). We denote
the optimal demand rates for the centralized system under the two types by λ∗A1 = λA1 (cS )
and λ̄∗A1 = λ̄A1 (cS ), respectively. We can take them as two quantity breakpoints. Accord-
ing to Proposition 2, if the production cost of the manufacturer is the common knowledge,
by inserting cM = cM into w1∗ , we obtain the unit wholesale price w ∗1A1 for the type-cM

manufacturer; similarly, we can obtain the unit wholesale price w̄1A 1
for the type-c̄M man-
ufacturer. To coordinate the channel, a quantity discount contract has to prevent one type
of the manufacturer from choosing the optimal breakpoint and wholesale price designed for
the other type of the manufacturer. Given a unit wholesale price wA1 , according to Proposi-
tion 1, we can rewrite the profit (4) of the manufacturer with the unit production cost cM as
πMA1 (cM , λA1 (wA1 ), wA1 ).
Proposition 3 summarizes the coordination mechanism.

Proposition 3 Under asymmetric information on the manufacturer’s unit production cost,


the all-unit quantity-discount coordination mechanism
⎧ ∗ ∗ ∗
⎨min{w1A1 , ŵA1 }, if λA1 ≥ λA1

wA∗ 1 = w̄1A

, if λ̄∗A1 ≤ λA1 < λ∗A1

⎩ ∗
1
w2A1 , if λA1 < λ̄∗A1
can coordinate the channel if and only if
  

πMA1 c̄M , λ̄∗A1 , w̄1A

1
≥ πMA1 c̄M , λ∗A1 , min w∗1A1 , ŵ∗A1 , (6)

where w2A 1
is sufficiently large such that the manufacturer does not choose it, λLH =
∗ ∗
max{λ̄A1 , λA1 (w̄1A 1
)} and
486 Ann Oper Res (2016) 241:475–496

a(s) − λ∗A1 kβ
ŵ∗A1 = − − cM
α α(μ − λ∗A1 )

λLH kβ a(s) − λLH ∗
+ − + c + w̄1A1 .
λ∗A1 α(μ − λLH ) α M

Comparing Propositions 2 and 3, we know that under the asymmetric information, to


induce the type-cM manufacturer to choose the right quantity breakpoint, the supplier may
decrease the unit wholesale price subject to the case with complete information, which is an
information rent to be paid by the supplier. In the above scheme, (6) ensures that the type-
c̄M manufacturer has no incentive to deviate its quantity breakpoint. If condition (6) is not
satisfied, the supply chain cannot be coordinated by an all-unit quantity discount contract
because the contract is not able to differentiate the two types of the manufacturer. This con-
dition shows that an all-unit quantity discount contract under complete information may not
be able to coordinate the channel for all cases. In particular, we have min{w∗1A1 , ŵ∗A1 } = w∗1A1
if and only if πMA1 (cM , λ∗A1 , w ∗1A1 ) ≥ πMA1 (cM , λLH , w̄1A

1
). This implies that, the type-cM
manufacturer may obtain a higher profit allocation than t · πC∗ because the supplier needs
to decrease the unit wholesale price from w∗1A1 to ŵ∗A1 to prevent the type-cM manufacturer
from deviating the right breakpoint. In other words, if we have a desired profit allocation
ratio t , we may have to use another different value t , t > t for the type-cM manufacturer in
the calculation of the contract design. When t − t is too large, (6) may not be satisfied, and
the channel cannot be coordinated by the above quantity discount contract.
In our numerical experiments, fortunately, we find that condition (6) can be satisfied for
virtually all cases with the same default setting in Table 1, showing that the quantity discount
contract is at least realistic in practice. Some numerical results are shown in Fig. 1, where
we report how the unit production costs for two types of the manufacturer may influence the
unit wholesale price.
From Fig. 1, we know that the unit wholesale price min{w∗1A1 , ŵ ∗A1 } increases with the
unit production cost cM while decreases with the unit production cost c̄M , which means that
the difference between the unit production costs for two types of the manufacturer (or vari-
ance of the manufacturer’s unit production cost) has a negative effect on min{w∗1A1 , ŵ∗A1 }.

Fig. 1 Unit wholesale price versus the manufacturer’s unit production costs
Ann Oper Res (2016) 241:475–496 487

Specifically, when the unit production cost cM increases, the cost difference/variance de-
creases such that the information rent for the type-cM manufacturer decreases, which re-
sults in a higher unit wholesale price for it, differing from that under complete information.
When the unit production cost c̄M increases, the cost difference/variance increases such that
the unit wholesale price min{w∗1A1 , ŵ ∗A1 } decreases due to a higher information rent for the

type-cM manufacturer and a lower unit wholesale price w̄1A 1
.

5.2 Endogenization of lead-time standard and channel coordination

In the basic model, we assume that the lead-time standard is endogenous. It is unclear how
the endogenization of the lead-time standard affects the coordination mechanism. This sub-
section addresses this issue by developing an alternative model with an exogenous lead-time
standard and comparing it with the basic model.
Note that the lead-time standard is exogenous now. Similar to Proposition 1, we can
show that when the delivery reliability constraint (1) is ignored, the optimal retail price of
the centralized system is pA∗ 2 = (cM + cS )/2 + (a(s) − βL)/(2α) and the optimal demand
rate is λ∗A2 = [a(s) − βL − α(cM + cS )]/2. Similar to the basic model, we assume that a0
is sufficiently large such that the demand rate λ∗A2 is positive. Then the manufacturer’s order
quantity (demand rate) reaction to the unit wholesale price is λA2 (wA2 ) = [a(s) − βL −
α(cM + wA2 )]/2. From (1), it follows that the demand rate should be less than μ − k/L.
Considering the delivery reliability constraint (1), the optimal demand rate (quantity break-
point) of the centralized system is λ∗∗ ∗
A2 = min{λA2 , μ − k/L}, and the corresponding optimal

retail price is max{pA2 , p̂A2 }, where p̂A2 = [k + L(a(s) − βL − μ)]/(αL). If the order quan-
tity reaction λA2 (wA2 ) is higher than μ − k/L, the quantity discount cannot come into effect
because it is unnecessary to stimulate the demand rate due to the delivery reliability con-
straint. Thus, the coordinated unit wholesale price wA2 should satisfy λA2 (wA2 ) < μ − k/L
if the delivery reliability constraint is binding (λ∗A2 ≥ μ − k/L). If the delivery reliability
constraint (1) under the centralized setting is nonbinding (λ∗A2 < μ − k/L), then the order
quantity reaction λA2 (wA2 ) is lower than μ − k/L because wA2 ≥ cS , which implies that
constraint (1) can be removed.
If the delivery reliability constraint (1) is binding (λ∗A2 ≥ μ − k/L), we will have the
following all-unit quantity discount mechanism



w1A , if λA2 ≥ μ − k/L
wA2 = ∗
2
w2A 2
, otherwise,
∗ ∗
where w1A 2
= cS t + (1 − t)(p̂A2 − cM ), w2A 2
is sufficiently high such that the manufacturer
L(2μ+α p̂ +βL−a(s))−2k
will not take it, and 0 ≤ t < t¯ = A2
αL(p̂A2 −cM −cS )
. That is, the quantity discount plays
a role in inducing a higher demand rate only when the manufacturer’s bargaining power
(t) is sufficiently low; otherwise, no quantity discount mechanism is needed because the
manufacturer will automatically choose the optimal retail price of the centralized system.
Similarly, if the delivery reliability constraint (1) is nonbinding (λ∗A2 < μ − k/L), the all-
unit quantity discount coordination mechanism is



w1A , if λA2 ≥ λ∗A2
wA2 = ∗
2
w2A2 , otherwise,

where w1A 2
= cS t + (1 − t)(pA∗ 2 − cM ) and w2A

2
is sufficiently large such that it is not
chosen.
488 Ann Oper Res (2016) 241:475–496

When the delivery reliability constraint (1) is binding under the centralized setting, the
quantity breakpoint of the coordination mechanism is independent of the costs (cM , cS ), and
sensitivity of demand to price (α), delivery reliability (γ ), and lead-time (β) because of the
capacity constraint, which results in that the retail price is independent of the costs. When
delivery reliability constraint (1) is nonbinding, the coordination mechanism is independent
of the capacity because the objectives of the manufacturer and the centralized system are
independent of it.
Proposition 4 summarizes some analytical results on comparative statics of the coordina-
tion mechanism, contrasting with the basic model (see Proposition 1 and Table 1).

Proposition 4 (i) When the delivery reliability constraint (1) is binding, the coordinated unit

wholesale price (w1A 2
) decreases with the capacity and lead-time standard; the quantity
breakpoint (λ∗∗A2 ) increases with the lead-time standard while decreases with the delivery
reliability standard;
(ii) When the delivery reliability constraint (1) is nonbinding, the coordinated unit whole-
sale price decreases with the lead-time standard; and the quantity breakpoint decreases with
the lead-time standard.

From Table 1, we know that when the lead-time standard is endogenous, a higher lead-
time standard goes often along with a higher coordinated unit wholesale price. However,
when the lead-time standard is exogenous, the coordinated unit wholesale price no longer
increases with the lead-time standard. It is now the case where the lead-time standard de-
creases the coordinated unit wholesale price. The effect of the interaction between the retail
price and the lead-time standard on demand rate accounts for this difference. Under the en-
dogenous lead-time setting, the manufacturer adjusts the retail price and lead-time to react
to the change of environment, which further affects the coordinated unit wholesale price as
well as quantity breakpoint. However, under the exogenous lead-time setting, a longer lead-
time standard decreases the demand rate. Thus, the manufacturer must decrease the retail
price to offset a part of the negative effect of the longer lead-time standard on the demand
rate.
Under the exogenous lead-time setting, when the lead-time standard increases, the quan-
tity breakpoint decreases due to its negative effect on the demand rate if the delivery reli-
ability constraint (1) is nonbinding, which is inconsistent with that under the endogenous
lead-time setting (see Table 1). However, if the delivery reliability constraint (1) is binding,
the quantity breakpoint increases with the lead-time standard due to the relaxation of the
constraint; and decreases with the delivery reliability standard because this intensifies the
conflict between the demand rate and capacity.

5.3 Endogenization of delivery reliability standard and channel coordination

We now consider the case where the delivery reliability standard s is an endogenous decision
variable of the manufacturer. Here, different delivery reliability standards s across firm mean
that the firms have different probabilities of delivering on time within LA3 time units. Thus,
we can rewrite (2) as
λ A 3 = a 0 + γ · s A 3 − α · pA 3 − β · L A 3 . (7)
Now the manufacturer determines the resale price pA3 , the leadtime standard LA3 and
the delivery reliability standard sA3 to maximize the long-run average profit per unit time
Ann Oper Res (2016) 241:475–496 489

πMA3 (pA3 , LA3 , sA3 ; wA3 ) = λA3 · (pA3 − cM − wA3 ) by solving the following optimization
problem.

max πMA3 (pA3 , LA3 , sA3 ; wA3 ), (8)


pA3 ,LA3 ,sA3

s.t. (μ − λA3 )LA3 ≥ kA3 . (9)


To solve the optimization problem, we introduce an assumption β < γ μ which is needed
for the existence of an optimal solution. Details can be referred to the proof of the following
Proposition 5. The business meaning of β < γ μ is that the demand rate is not too sensi-
tive to the delivery leadtime. Then it can be shown that the optimal quantity (demand rate)
λA3 (wA3 ) is the optimal solution of
 
max πMA3 pA3 (wA3 ), LA3 (wA3 ), sA3 (wA3 ); wA3 ,
{λA3 |f2 (wA3 ,λA3 )=0
and 0<λA3 <μ}

where f2 (wA3 , λA3 ) = (μ − λA3 )[(μ − λA3 )(a0 − αcM − αwA3 + γ − 2λA3 ) − β] −
βμ ln[1/(1 − sA3 )]. We have Proposition 5 to summarize the optimal solution of (8).

Proposition 5 Assume β < γ μ. Given the unit wholesale price wA3 , the manufacturer’s
optimal delivery reliability standard is sA3 (wA3 ) = 1 − β/[γ (μ − λA3 (wA3 ))], optimal lead-
time standard is LA3 (wA3 ) = ln[1/(1 − sA3 (wA3 ))]/(μ − λA3 (wA3 )), and the optimal retail
price is pA3 (wA3 ) = (a0 − β · LA3 (wA3 ) + γ · sA3 (wA3 ) − λA3 (wA3 ))/α.

Proposition 5 implies that the equilibrium demand rate λA3 (wA3 ) is less than (a0 − αcM −
αwA3 + γ )/2. We also have the following results on comparative statics.

Corollary 4 λA3 (wA3 ) is an increasing function of γ if and only if the manufacturer’s ca-
pacity satisfies μ > μ̂ = λA3 (wA3 ) + LA1 ln(1 + LA3 λA3 (wA3 )); λA3 (wA3 ) is always decreas-
3
ing with α, β, cM and wA3 while increasing with μ.

Corollary 4 implies that, when the capacity μ is sufficiently low (μ < μ̂), the direct
positive effect of delivery reliability sensitivity γ on demand rate is weaker than its indirect
negative effect through resale price and service level constraint, which implies that λA3 (wA3 )
is a decreasing function of γ . Corollary 4 means that if λA3 (wA3 ) is an increasing function
of γ , then the manufacturer’s capacity μ is higher than the threshold μ̂.
From Programming (8) and Proposition 5, we know that the optimal demand rate (real-
ized quantity) of the centralized channel is λ∗A3 = λA3 (cS ) and the maximum profit of the

centralized channel is πCA 3
= πMA3 (pA3 (cS ), LA3 (cS ), sA3 (cS ); cS ). We can take λ∗A3 as the
quantity breakpoint.
Similar to Proposition 2, we can show that the all-unit quantity discount mechanism
 ∗
∗ w1A3 , if λA3 ≥ λ∗A3
wA3 = ∗
w2A 3
, otherwise
∗ ∗
can coordinate the channel, where w1A 3
= cS · t + (1 − t)(pA3 (cS ) − cM ), w2A 3
> ŵ2A3
∗ ∗
(> w1A 3
) with ŵ2A3 satisfying π MA3 (p A3 (ŵ2A3 ), L A3 (ŵ2A3 ), s A3 (ŵ2A3 ); ŵ2A3 = t · πCA3 .
)
From Proposition 1 and Corollary 4, we know that the monotonicity of the quantity break-
point (λ∗A3 ) in the price sensitivity (α), lead-time sensitivity (β), and the costs (cM and cS )
is similar to that under the exogenous delivery reliability setting (basic model). It is worthy
to note that under the exogenous delivery reliability setting, the quantity breakpoint (λ1 )
490 Ann Oper Res (2016) 241:475–496

increases with the delivery reliability sensitivity (γ ) due to a higher market scale. How-
ever, under the endogenous delivery reliability setting, the quantity breakpoint (λ∗A3 ) may
not increase with it. It is now the case that the quantity breakpoint decreases with it if the
capacity is sufficiently small (μ < μ̂). This will further affect the coordinated unit wholesale
price. The cross effect between the delivery reliability sensitivity and the demand rate on the
manufacturer’s profit accounts for this difference.

5.4 Endogenization of capacity and channel coordination

Up to now, we assume that the manufacturer’s capacity is predetermined, i.e., the capacity
has been fixed before the supplier offers the contract to the manufacturer. However, some-
times, owing to the contract arrangement represents a long-term relationship between the
supplier and the manufacturer, the supplier must offer the contract before the capacity de-
cision is made. In this case, how to coordinate the manufacturer’s decisions (especially,
capacity)? A higher capacity can better satisfy the consumer demand by reducing lead-time,
which will result in a higher demand rate. However, raising capacity incurs an extra cost.
Let the unit capacity cost be cμ . The manufacturer will make a trade-off between the benefit
and cost of increasing capacity. Similar to (3), we can give the demand rate λA4 . From (1),
it follows that the optimal capacity is μA4 = λA4 + k/LA4 .
Similar to (4), the central decision-maker chooses

πCA4 (LA4 , pA4 ) = λA4 · (pA4 − cM − cS ) − cμ · (λA4 + k/LA4 )


= λA4 · (pA4 − cM − cS − cμ ) − cμ · k/LA4 . (10)
The term cμ · k/LA4 is the extra capacity cost dealt with uncertainty.

To show the optimal solution to problem (10), we define λ̂A4 = 3 cμ kαβ/24/3 , f3 (λA4 ) =
  
− cμ kαβ + [a(s) − α(cμ + cM + cS )] λA4 − 2( λA4 )3 , and λ∗A4 as the unique root of the
first-order condition f3 (λA4 ) = 0 over the interval [λ̂A4 , +∞). We further assume f3 (λ̂A4 ) ≥
0 which ensures that the solution (λ∗A4 ) of the equation f3 (λA4 ) = 0 exists; details can be
referred to the proof of Proposition 6. Then we have

Proposition 6 When the capacity is endogenous, the optimal  decisions of the centralized
system are as follows: the optimal lead-time standard is LA4 = cμ kα/(βλ∗A4 ), the optimal

retail price is pA∗ 4 = (a(s) − βL∗A4 − λ∗A4 )/α and the optimal capacity is μ∗A4 = λ∗A4 + k/L∗A4 .

From Proposition 6, we know that the optimal decisions of the centralized system can
be represented by the optimal demand rate. A higher demand rate is along with a shorter
lead-time and larger capacity. That is, the central decision-maker would like to reduce the
lead-time to stimulate demand rate and increase capacity to meet the delivery reliability
constraint. However, whether a higher demand rate goes along with a lower retail price
depends on the interaction between the retail price and the lead-time.
Proposition 7 addresses how to coordinate the supply chain with an endogenous capacity.

Proposition 7 The all-unit quantity discount mechanism





w1A , if λA4 ≥ λ∗A4
wA4 = ∗
4
w2A 4
, otherwise,
coordinates the supplier-manufacturer channel in which wA∗ 4 depends on the order quan-

tity (realized demand rate) λA4 , where w1A 4
= cS · t + (1 − t)(pA∗ 4 − cM − cμ ) − cμ k(1 −
Ann Oper Res (2016) 241:475–496 491

t)/(L∗A4 λ∗A4 ), and w2A



4
is sufficiently high such that the manufacturer has no incentive to
take it.

It is counterintuitive that the all-unit quantity discount scheme can still coordinate the
supply chain with endogenous price, leadtime standard and capacity. One expects that a new
dimension is needed to add into the scheme because one more (capacity) decision needs
to be coordinated. However, although there are three decision variables, the delivery relia-
bility constraint (1) helps eliminate one decision dimension. Thus, in essence, the supplier
needs to design a mechanism to coordinate only two decisions. Similar to Proposition 6,
we can show that the unit wholesale price of the supplier affects the manufacturer’s deci-
sions through the quantity (demand rate) reaction of the manufacturer due to the absence
of a cross term including the product of the unit wholesale price and leadtime standard.
An all-unit quantity discount scheme is an effective tool to induce the optimal demand rate
(quantity breakpoint) of the centralized system. Thus, the supplier can still coordinate the
price, leadtime and capacity decisions via an all-unit quantity discount mechanism. Unlike
the exogenous capacity setting, the supplier should bear a part of the extra capacity cost to
deal with uncertainty.
Similar to Corollary 1, we can show that the quantity breakpoint λ∗A4 decreases with
the costs (cμ , cM , cS ), the price sensitivity (α), and the lead-time sensitivity (β), while in-
creases with the basic market scale (a0 ) and the delivery reliability sensitivity (γ ), which
are consistent with those in the basic model. That is, the monotonicity of the quantity break-
point is robust to the endogenization of capacity. Under the exogenous capacity setting, the
coordination mechanism can ignore the capacity investment because it does not affect the
manufacturer’s decisions. However, under the endogenous capacity setting, ignoring the ca-
pacity investment will result in the absence of coordination because the manufacturer has
an incentive to deviate from the optimal capacity of the centralized system, which further
affects the other decisions and results in system inefficiency.

6 Conclusions

This paper considers a two-stage supply chain with demand sensitive to multiple factors,
namely the selling price, the announced delivery time, and the reliability of satisfying the
delivery time announcement. Comparing with the literature on leadtime- and price-sensitive
demand, the new contribution of our work is to explicitly model the impact of the delivery
reliability standard on the demand rate. We focus on how to coordinate the manufacturer’s
decisions and on the robustness of coordination mechanism under various scenarios.
We investigate the coordination of the channel via an all-unit quantity discount contract
under different scenarios. Besides the coordination contract design, we also focus on some
managerial insights of the channel, for example, how the delivery reliability standard may
affect the demand rate and the channel profit, how the price and leadtime quotations may
complement or substitute with each other, how the endogenization of delivery reliability
standard and/or leadtime standard affects the coordination mechanism, how the difference
between the unit production costs (cost variance) for two types of the manufacturer influ-
ences the coordination mechanism, and whether the all-unit quantity discount scheme can
still coordinate the supply chain when the capacity, leadtime standard and price are endoge-
nous. Such insights will help both the supplier and the manufacturer to make the optimal
operations and marketing decisions.
492 Ann Oper Res (2016) 241:475–496

We suggest the following future research opportunities. First, other coordination schemes
are needed to address the case of information asymmetry so that an unconditional coordina-
tion can be achieved. Second, an information revelation mechanism may be designed by the
supplier under the asymmetric information setting. Finally, a model with one supplier and
multiple manufacturers may be more close to practice.

Acknowledgements We would like to thank the two anonymous referees and the guest editors for their
many helpful suggestions and insightful comments, which have significantly improved the presentation of
this paper. This research was supported in part by the National Natural Science Foundation of China under
Grants 70971060 and 70731002, and by Hong Kong RGC GRF 618807.

Appendix

Proof of Corollary 1 Differentiating f1 (w, λ(w)) with respect to w and λ, we have


   2
∂f1 w, λ(w) /∂w = −α μ − λ(w) < 0,

     
∂f1 w, λ(w) /∂λ = −2 μ − λ(w) a(s) − α(cM + w) − 2λ(w)
 

+ μ − λ(w) < 0, (11)


which follows from λ(w) < μ and a(s) − α · (cM + w) − 2λ(w) > 0. Furthermore, we have
dλ(w) ∂f1 (w, λ(w))/∂w
=− < 0.
dw ∂f1 (w, λ(w))/∂λ
Furthermore, from Proposition 1, it follows that L(w) is a decreasing function of w, and
p(w) is an increasing function of w. 

Proof of Corollary 2 From (11), it follows that


     
dλ(w) ∂f1 (w, λ(w))/∂s ∂f1 (w, λ(w))
sign = sign − = sign .
ds ∂f1 (w, λ(w))/∂λ ∂s
Differentiating f1 (w, λ(w)) with respect to s, we have
   2
∂f1 w, λ(w) /∂s = μ − λ(w) · γ − βμ/(1 − s). (12)

Thus, λ(w) is an increasing function of s if and only if λ(w) ≤ μ − βμ/[γ (1 − s)]. Thus,
Corollary 2 follows because of λ(w) < μ. 

Proof of Proposition 2 From Proposition 1, it follows that, to coordinate the supply chain,
we only need to design a mechanism to induce the demand rate in the centralized setting λ1 .
Note that λ(w) is a decreasing function of w and w ≥ cS . If the manufacturer would like
to access the unit wholesale price w1∗ , its optimal quantity is λ1 . When the supply chain is
coordinated, since the supplier wants to achieve a fraction (1 − t) of the maximum channel
profit πC∗ , the manufacturer’s profit is

  a(s) − λ1 kβ
πM L(cS ), p(cS ); w1∗ = tπC∗ = tλ1 − − cM − cS . (13)
α α(μ − λ1 )
Solving (13) for w1∗ , we have
   
w1∗ = a(s) − λ1 (1 − t)/α + cS t − cM (1 − t) − kβ(1 − t)/ α(μ − λ1 ) ,
Ann Oper Res (2016) 241:475–496 493

which is greater than cS following from 0 < t < 1 and πC∗ > 0, i.e., w1∗ > cS .
When the manufacturer wants to access the unit wholesale price w = w2∗ , from Proposi-
tion 1, we know that the manufacturer will offer p(w2∗ ) and L(w2∗ ) to maximize its profit.
Thus, its optimal profit is
 
  a(s) − λ(w) kβ
πM L(w), p(w); w = λ(w) − − cM − w ,
α α(μ − λ(w))
with w = w2∗ . (14)
From (5), it follows that dπM (L(w), p(w); w)/dw = −λ(w) < 0. Thus, since λ(w) is a
decreasing function of w, πM (L(w), p(w); w) is a decreasing and convex function of w. Let
the unique root of πM (L(w), p(w); w) = tπC∗ be ŵ2 (≥ cS ). Furthermore, the manufacturer
would like to choose w1∗ if w2∗ > ŵ2 . From w1∗ > cS and Proposition 1, it follows that
         
πM L w1∗ , p w1∗ ; w1∗ > πM L(cS ), p(cS ); w1∗ = tπC∗ = πM L(ŵ2 ), p(ŵ2 ); ŵ2 ,
which implies w1∗ < ŵ2 , i.e., w∗ is a quantity discount scheme. 

Proof of Corollary 3 Under the all-unit quantity discount mechanism w∗ , the equilibrium
profit of the manufacturer is πM (L(cS ), p(cS ); w1∗ ) = tπC∗ . From Proposition 1 and (13), it
follows that ∂πC∗ /∂λ1 = 0. Furthermore, differentiating πC∗ with respect to s, we have

dπC∗ ∂πC∗ dλ1 ∂πC∗ ∂πC∗ λ1 β
= + γ+ = γ− .
ds ∂λ1 ds ∂a ∂s α (μ − λ1 )(1 − s)
Similarly, we can show the others. 

Proof of Proposition 3 Under the case with two types of the manufacturer, we consider
the all-unit quantity discount scheme with the quantity breakpoints λA∗ 1 and λ̄∗A1 and the unit
wholesale prices corresponding to three intervals, where λA∗ 1 > λ̄∗A1 .
In order to assure that each type of the manufacturer chooses the “right” breakpoint, the
coordination mechanism should prevent the type-cM manufacturer from ordering a quantity
λA1 < λ∗A1 , and prevent the type-c̄M manufacturer from ordering a quantity λA1 ≥ λ∗A1 . In
other words, the following inequalities should hold under the mechanism:
     
πMA1 c̄M , λ̄∗A1 , w̄1A

1

≥ πMA1 c̄M , λA1 , w̄1A 1
, for λA1 ∈ λ̄∗A1 , λ∗A1 , (15)
  

πMA1 c̄M , λ̄∗A1 , w̄1A

1
≥ πMA1 c̄M , λA1 , min w∗1A1 , ŵ∗A1 , for λA1 ≥ λ∗A1 , (16)

 

πMA1 cM , λ∗A1 , min w∗1A1 , ŵ∗A1 ≥ πMA1 cM , λA1 , min w ∗1A1 , ŵ∗A1 ,
for λA1 > λ∗A1 , (17)

    ∗ 
πMA1 cM , λ∗A1 , min w∗1A1 , ŵ∗A1 ≥ πMA1 cM , λA1 , w̄1A

1
, for λA1 ∈ λ̄A1 , λ∗A1 . (18)
Since the optimal demand rate of the type-c̄M manufacturer λ̄A1 (wA1 ) is a decreasing
∗ ∗
function of wA1 and w̄1A 1
> cS , we have λ̄A1 (w̄1A 1
) < λ̄A1 (cS ) = λ̄∗A1 . Thus, it follows

from the concavity of πMA1 over λA1 that πMA1 (c̄M , λA1 , w̄1A 1
) decreases with λA1 for
∗ ∗
λA1 ∈ (λ̄A1 , λA1 ), which implies that condition (15) is satisfied. Similarly, we can show con-
dition (17).
∗ ∗
Corollary 1 and w̄1A 1
> cS imply that, when the unit wholesale price is w̄1A 1
, the
∗ ∗
optimal demand rate of the type-cM manufacturer satisfies λA1 (w̄1A1 ) < λA1 . Note that
∗ ∗
πMA1 (cM , λA1 , w̄1A 1
) is a concave function of λA1 . Thus, we have πMA1 (cM , λA1 , w̄1A 1
)≤
∗ ∗ ∗ ∗ ∗
πMA1 (cM , λLH , w̄1A1 ) for λA1 ∈ [λ̄A1 , λA1 ), where λLH = max{λ̄A1 , λA1 (w̄1A1 )}.
494 Ann Oper Res (2016) 241:475–496

Let the solution of πMA1 (cM , λ∗A1 , wA1 ) = πMA1 (cM , λLH , w̄1A ∗
1
) for wA1 be ŵ∗A1 , given

in Proposition 3. Furthermore, since πMA1 (cM , λA1 , wA1 ) is a decreasing function of wA1 ,
we have πMA1 (cM , λ∗A1 , min{w∗1A1 , ŵ∗A1 }) ≥ πMA1 (cM , λ∗A1 , ŵ∗A1 ) = πMA1 (cM , λLH , w̄1A ∗
1
)≥
∗ ∗ ∗
πMA1 (cM , λA1 , w̄1A1 ), for λA1 ∈ [λ̄A1 , λA1 ), i.e., condition (18) is always satisfied.
Corollary 1 and min{w∗1A1 , ŵ∗A1 } > cS imply λ̄A1 (min{w∗1A1 , ŵ∗A1 }) < λ̄∗A1 < λ∗A1 . More-
over, note that πMA1 (c̄M , λA1 , min{w∗1A1 , ŵ∗A1 }) is a concave function of λA1 . Thus,
πMA1 (c̄M , λA1 , min{w∗1A1 , ŵ∗A1 }) is a decreasing function of λA1 and πMA1 (c̄M , λ∗A1 ,
min{w∗1A1 , ŵ ∗A1 }) ≥ πMA1 (c̄M , λA1 , min{w∗1A1 , ŵ ∗A1 }) for λA1 ≥ λ∗A1 . Furthermore, condition
(16) is satisfied if and only if πMA1 (c̄M , λ̄∗A1 , w̄1A

1
) ≥ πMA1 (c̄M , λ∗A1 , min{w∗1A1 , ŵ ∗A1 }).
Suppose min{w∗1A1 , ŵ∗A1 } ≥ w̄1A ∗
1
. Note that πMA1 (cM , λA1 , wA1 ) is a decreasing func-
tion of wA1 and a concave function of λA1 . Thus, it follows from λA1 (w̄1A ∗
1
) < λ∗A1 that
∗ ∗ ∗ ∗ ∗ ∗
πMA1 (cM , λA1 , w̄1A1 ) > πMA1 (cM , λA1 , w̄1A1 ) ≥ πMA1 (cM , λA1 , min{w1A1 , ŵ A1 }), for λA1 ∈
[λLH , λ∗A1 ) i.e., the manufacturer with cM has an incentive to order a smaller quantity than
λ∗A1 to access a lower unit wholesale price w̄1A ∗
1
, contradicting (18). Thus, we analytically
∗ ∗ ∗ ∗
show min{w1A1 , ŵA1 } < w̄1A1 , i.e., wA1 is a quantity discount mechanism. 

Proof of Proposition 5 Similar to Pekgün et al. (2008), we can substitute λA3 for pA3 as
a decision variable. From (7), we have pA3 = (a0 − β · LA3 + γ · sA3 − λA3 )/α. Note that
both sA3 and λA3 are the decision variables and we can not regard λA3 as a function of sA3
again. Inserting pA3 into (8), we know that πMA3 is an increasing function of sA3 . Further, (9)
implies that the manufacturer increases the delivery reliability standard sA3 until the delivery
reliability constraint is binding, i.e., ln 1−s1A = LA3 (μ − λA3 ) or sA3 = 1 − e−LA3 (μ−λA3 ) .
3
Furthermore, Programming (8) is equivalent to determining LA3 and λA3 to maximize
 
πMA3 (LA3 , λA3 ) = λA3 a0 − βLA3 + γ − γ · e−LA3 (μ−λA3 ) − λA3 − α(cM + wA3 ) /α.
(19)

The first-order conditions of (19) are


 
∂πMA3 /∂LA3 = (λA3 /α) · γ (μ − λA3 )e−LA3 (μ−λA3 ) − β = 0, (20)


∂πMA3 /∂λA3 = a0 − α · (cM + wA3 ) − β · LA3 + γ − γ · e−LA3 (μ−λA3 ) (1 + LA3 λA3 )

− 2λA3 /α = 0. (21)

Solving (20) for LA3 , we have


 
LA3 (λA3 ) = ln γ (μ − λA3 )/β /(μ − λA3 ).

From β < γ μ, it follows that there exists a positive demand rate λA3 satisfying LA3 (λA3 ) >
0. Using LA3 (λA3 ), we find that the first-order condition (21) is equivalent to
 
f2 (wA3 , λA3 ) = (μ − λA3 ) (μ − λA3 )(a0 − αcM − αwA3 + γ − 2λA3 ) − β
 
− βμ ln γ (μ − λA3 )/β = 0.

Note that the manufacturer will achieve a zero profit if λA3 = 0. Moreover, the delivery
reliability constraint (9) implies λA3 < μ. Thus, the optimal demand rate satisfies the first-
order condition f2 (wA3 , λA3 ) = 0. 
Ann Oper Res (2016) 241:475–496 495

Proof of Corollary 4 From (19), we have


∂ 2 πMA3 (LA3 , λA3 )  
= 1 − e−LA3 (μ−λA3 ) (1 + LA3 λA3 ) /α,
∂λA3 ∂γ

which is positive if and only if μ > μ̂. Note ∂ 2 πMA3 (LA3 (wA3 ), λA3 (wA3 ))/∂λ2A3 < 0. Ac-
cording to implicit function theorem, we have
   2   2 
∂λA3 (wA3 ) ∂ πMA3  ∂ 2 πMA3 ∂ πMA3
sign = sign − = sign .
∂γ ∂λA3 ∂γ ∂λ2A3 ∂λA3 ∂γ

Further, λA3 (wA3 ) is an increasing function of γ if and only if μ > μ̂. Similarly, we can
show the others. 

Proof of Proposition 6 Similar to Pekgün et al. (2008), we can substitute λA4 for pA4 as a
decision variable. From (3), we obtain pA4 = (a(s) − βLA4 − λA4 )/α. Thus, we can rewrite
profit function (10) as
  
πCA4 (LA4 , λA4 ) = λA4 · a(s) − βLA4 − λA4 /α − cM − cS − cμ − cμ · k/LA4 . (22)

The first-order conditions of (22) are

∂πCA4 (LA4 , λA4 )/∂LA4 = cμ k/L2A4 − βλA4 /α = 0, (23)


 
∂πCA4 (LA4 , λA4 )/∂λA4 = a(s) − LA4 β − 2λA4 /α − (cμ + cM + cS ) = 0. (24)

Solving (23) for LA4 , we have LA4 (λA4 ) = cμ kα/(βλA4 ). Using LA4 (λA4 ), (24) is equiv-
alent to
   
f3 (λA4 ) = − cμ kαβ + a(s) − α(cμ + cM + cS ) λA4 − 2( λA4 )3 = 0. (25)

Here f3 (λA4 ) is a convex-concave function of λA4 . Note that when the central decision-
maker offers the price pA4 = cμ + cM + cS , the demand rate (3) shouldbe positive, i.e.,
a(s) − α(cμ + cM + cS ) > βLA4 . Moreover, (25) implies f3 (0) = − cμ kαβ < 0 and
f3 (+∞) < 0. If f3 (λ̂A4 ) ≥ 0, there exist two positive roots for (25). Denote the max-
imum root as λ∗A4 . We can show that πCA4 (LA4 , λA4 ) is concave over (LA4 , λA4 ) only
 
when LA4 < 3 4cμ kα/β 2 . Using LA4 (λA4 ), LA4 < 3 4cμ kα/β 2 is equivalent to λA4 > λ̂A4 =

3
cμ kαβ/24/3 . From f3 (λ̂A4 ) ≥ 0 and f3 (+∞) < 0, we see the maximum root λ∗A4 ≥ λ̂A4 .
Further, the optimal demand rate is λ∗A4 . The other root minimizes the profit of the central-
ized system. 

Proof of Proposition 7 Under the decentralized supply chain, similar to Proposition 6,


we can obtain the optimal reactions of the manufacturer by substituting the unit wholesale
price wA4 for the supplier’s unit production cost cS . Denote the manufacturer’s optimal
demand rate reaction to the unit wholesale price as λA4 (wA4 ). Similar to Corollary 1, we can
show that λA4 (wA4 ) decreases with wA4 . It is obvious that the unit wholesale price should
∗ ∗
satisfy w1A 4
> cS . From w1A 4
> cS and Proposition 6, it follows that λA4 (wA4 ) is lower
than λA4 , i.e., λA4 (w1A4 ) < λ∗A4 . Thus, if the manufacturer wants to take w1A
∗ ∗ ∗
4
, the optimal
∗ ∗
quantity (realized demand rate) is λA4 . Similar to Proposition 2, we can obtain w1A 4
, given
in Proposition 7. 
496 Ann Oper Res (2016) 241:475–496

References

Allon, G., & Federgruen, A. (2007). Competition in service industries. Operations Research, 55(1), 37–55.
Benjaafar, S., Elahi, E., & Donohue, K. L. (2007). Outsourcing via service competition. Management Science,
53(2), 241–259.
Cachon, G. P. (2004). Supply chain coordination with contracts. In D. Simchi-Levi, S. D. Wu, & Z. J. Shen
(Eds.), Handbooks in operations research and management science: supply chain management (Vol. 11,
pp. 13–59). Amsterdam: Elsevier.
Chen, F., Federgruen, A., & Zheng, Y. S. (2001). Coordination mechanisms for a distribution system with
one supplier and multiple retailers. Management Science, 47(5), 693–708.
Corbett, C. J. (2001). Stochastic inventory systems in a supply chain with asymmetric information: cycle
stocks, safety stocks, and consignment stock. Operations Research, 49(4), 487–500.
Corbett, C., & de Groote, X. (2000). A supplier’s optimal quantity discount policy under asymmetric infor-
mation. Management Science, 46(3), 444–450.
Corbett, C. J., Zhou, D., & Tang, C. S. (2004). Designing supply contracts: contract type and information
asymmetry. Management Science, 50(4), 550–559.
Dana, J. (2001). Competition in price and availability when availability is unobservable. The Rand Journal of
Economics, 32(3), 497–513.
Desiraju, R., & Moorthy, S. (1997). Managing a distribution channel under asymmetric information with
performance requirements. Management Science, 43(12), 1628–1644.
Elhafsi, M. (2002). Optimal leadtimes planning in serial production systems with earliness and tardiness
costs. IIE Transactions, 34(3), 233–243.
Gilbert, S. M., & Weng, Z. K. (1998). Incentive effects favor nonconsolidating queues in a service system:
the principal-agent perspective. Management Science, 44(12), 1662–1669.
Gupta, D., & Benjaafar, S. (2004). Make-to-order, make-to-stock, or delay product differentiation? A com-
mon framework for modeling and analysis. IIE Transactions, 36(6), 529–546.
Ha, A. Y. (2001). Supplier-Buyer contracting: asymmetric cost information and cutoff level policy for buyer
participation. Naval Research Logistics, 48(1), 41–64.
Ha, A. Y., Li, L., & Ng, S. M. (2003). Price and delivery logistics competition in a supply chain. Management
Science, 49(9), 1139–1153.
Kolay, S., Shaffer, G., & Ordover, J. (2004). All-units discounts in retail contracts. Journal of Economics &
Management Strategy, 13(3), 429–459.
Liu, L., Parlar, M., & Zhu, S. X. (2007). Pricing and lead time decisions in decentralized supply chains.
Management Science, 53(5), 713–725.
Özer, Ö., & Wei, W. (2006). Strategic commitments for an optimal capacity decision under asymmetric
forecast information. Management Science, 52(8), 1238–1257.
Palaka, K., Erlebacher, S., & Kropp, D. (1998). Leadtime setting, capacity utilization, and pricing decisions
under leadtime dependent demand. IIE Transactions, 30(2), 151–163.
Pekgün, P., Griffin, P. M., & Keskinocak, P. (2008). Coordination of marketing and production for price and
leadtime decisions. IIE Transactions, 40(1), 12–30.
Rao, U. S., Swaminathan, J. M., & Zhang, J. (2005). Demand and production management with uniform
guaranteed leadtime. Production and Operations Management, 14(4), 1–13.
Shang, W., & Liu, L. (2011). Promised delivery time and capacity games in time-based competition. Man-
agement Science, 57(3), 599–610.
Song, J. S., Yano, C. A., & Lerssrisuriya, P. (2000). Contract assembly: dealing with combined supply lead
time and demand quantity uncertainty. Manufacturing & Service Operations Management, 2(3), 287–
296.
Upasani, A., & Uzsoy, R. (2008). Incorporating manufacturing lead times in joint production marketing
models: a review and some future directions. Annals of Operations Research, 161(1), 171–188.
Weng, Z. K. (1995). Channel coordination and quantity discounts. Management Science, 41(9), 1509–1522.
Xiao, T. J., Yang, D. Q., & Shen, H. C. (2011). Coordinating a supply chain with a quality assurance policy
via a revenue-sharing contract. International Journal of Production Research, 49(1), 99–120.
Yang, B., & Geunes, J. (2007). Inventory and lead time planning with leadtime-sensitive demand. IIE Trans-
actions, 39(5), 439–452.
Yano, C. A. (1987). Stochastic leadtimes in two-level assembly systems. IIE Transactions, 19(4), 371–378.
Zhang, T. L., Liang, L., Yu, Y. G., & Yu, Y. (2007). An integrated vendor-managed inventory model for a
two-echelon system with order cost reduction. International Journal of Production Economics, 109(1–
2), 241–253.
Reproduced with permission of the copyright owner. Further reproduction prohibited without
permission.

Anda mungkin juga menyukai