Omega 36 (2008) 838 – 851

www.elsevier.com/locate/omega

Vertical cost information sharing in a supply chain with

value-adding retailers夡
Dong-Qing Yaoa,∗ , Xiaohang Yueb , John Liuc
a Department of Management, College of Business and Economics, Towson University, MD 21252, USA
b School of Business Administration, University of Wisconsin-Milwaukee, WI 53201, USA
c Department of Logistics, Hong Kong Polytechnic University, Hung Hom, Kowloon, China

Received 12 August 2005; accepted 19 April 2006

Available online 9 June 2006

Abstract
We consider a supply chain consisting of one supplier and two value-adding heterogeneous retailers. Each retailer has full
knowledge about his own value-added cost structure that is unknown to the supplier and the other retailer. Assuming there is no
horizontal information sharing between two retailers, we model the supply chain with a three-stage game-theoretic framework. In
the first stage each retailer decides if he is willing to vertically disclose his private cost information to the supplier. In the second
stage, given the information he has about the retailers, the supplier announces the wholesale price to the retailers. In response to the
wholesale price, in the third stage, the retailers optimize their own retail prices and the values added to the product, respectively.
Under certain conditions, we prove the existence of equilibrium prices and added values. Furthermore, we obtain the condition under
which both retailers are unwilling to vertically share their private information with the supplier, as well as the conditions under
which both retailers have incentives to reveal their cost information to the supplier, thus leading to a win–win situation for the whole
supply chain.
䉷 2006 Elsevier Ltd. All rights reserved.

Keywords: Value-added; Supply chain; Information sharing; Distribution channel; Game theory

1. Introduction

The concept of “value-added”, the idea of adding service or components to a product to increase its value or price
[1], has been a buzzword in recent years. In order not to be disintermediated in this competitive E-business era, each
stage in a supply chain needs to add appropriate value to a product. There are numerous ways to add value to a product.
A retailer can bundle the product with value-added service/delivery or provide desirable value-added packaging [2].
In the IT industry, value can be added through services, free software, technical training or maintenance. For example,
software companies (e.g., Systemax) may bundle the PC with a package of free internet access such as basic AOL
service [3]. In hi-tech industries (e.g., electronics industry), value can be added through simple components labeling
and kitting to complex supply chain management service [4].

夡 This manuscript was processed by Associated Editor Richard Metters.

∗ Corresponding author. Tel.: +1 410 704 6185; fax: +1 410 704 3236.
E-mail addresses: dyao@towson.edu (D.-Q. Yao), xyue@uwm.edu (X. Yue), lgtjliu@polyu.edu.hk (J. Liu).

0305-0483/$ - see front matter 䉷 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.omega.2006.04.003
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 839

Since there are always operating costs incurred related to corresponding value-added service, one natural research
question arises: how much value should be added, especially under a competitive market structure? In this paper, we
will examine a distribution channel in a supply chain made up of one supplier and two heterogeneous value-adding
retailers/distributors. Both retailers order generic products from a supplier, and add certain value to the product, then
sell the product to customers. We are interested in studying how much appropriate value should be added to the generic
product by each retailer.
Many business practices around such types of distribution channels can be found in real life. For example, IT
solution providers may order white box PCs from suppliers, then add appropriate value (e.g. offer antivirus software,
technical training, provide extended warranty). Actually, CRN research showed that white boxes are their best-selling
desktops, compared to branded name PCs [5]. In the electronics industry, distributors may offer value-added services
such as connector and cable assemblies, customized integration, supply chain management services such as a vendor
managed inventory (VMI) program, or even design service [6]. In the franchising industry, each franchisee acquires
the franchisor’s generic product (the rights to use franchisor’s names, trademarks, etc.), then adds value and sells to the
customers [7]. In the retailing industry, retailers may add value to electronic products by providing an extended warranty
or offering lenient returns policy. For example, Best Buy’s warranty policy would cover defects the manufacturer does
not cover, while Wal-Mart’s warranty policy begins after the manufacturer’s policy expires [8].
Researchers have examined distribution channels with one player working with two other players. For example, Choi
[9] studied price competition in a distribution channel with two suppliers and one retailer. Under a different market
structure, he analytically obtained channel decisions for three non-cooperative games between the manufacturers and
the retailer. Tsay and Agrawal [10] investigated a supply chain with one supplier supplying a common product to
two retailers, and both retailers competing with each other along both price and service policies. Tsay and Agrawal
examined each party’s pricing strategy, total sales, market share and profitability.
However, existing studies of distribution channels assume supplier and retailers compete with complete informa-
tion. In other words, the research assumes that information such as production cost, operating cost and other market
parameters about supplier or retailers are common knowledge to all parties in a supply chain. This assumption may
not hold in real life. In practice, each party in a supply chain usually possesses private information that is unknown to
outsiders, and these firms are likely to protect their sales strategies by hiding their private cost or demand information
[11], thus leading to a competition under incomplete information. For example, warranty cost is confidential informa-
tion for retailers such as Best Buy and Circuit City because significant portions of their operating profits are from the
warranty service [8]. Franchisees also hold their private cost information of sale and decide if they will reveal their cost
information to the franchisor in the contract [7]. Therefore, what information to share and how to share information
becomes an interesting research issue.
Some researchers have studied information (e.g., demand, cost, and inventory) sharing/asymmetry in a supply chain.
For example, Li [12] studied the incentives a firm would need to share its private information (demand and costs) with
its competitors; Gavirneni [13] analyzed a supply chain where inventory information is shared between the supplier
and retailer. Cachon and Fisher [14] investigated the sharing of demand and inventory data in a supply chain with one
supplier and multiple identical retailers. Agrell et al. [15] examined information sharing in telecom supply chains where
a supplier has private cost information and investment opportunities. Corbett et al. [16] analyzed a supply chain with
one supplier and one buyer, and studied the value to the supplier of offering more general contracts and acquiring more
accurate information about the buyer’s cost structure. Assuming that information asymmetry between a manufacturer
and a retailer, i.e., the retailer’s knowledge of the manufacturer’s cost is incomplete, Lau et al. [17] also studied how to
set up the wholesale price and retail price in a supply chain.
In this paper, we will examine a supply chain consisting of one supplier and two heterogeneous retailers. There is
horizontal competition between the two retailers, each of whom orders common generic products from the supplier,
adds value to the products, then sells to customers. In addition, the supplier has no complete knowledge of the retailers’
cost structure regarding value-added information. Each retailer needs to decide if he is willing to vertically share his
cost information with the supplier.
Previously, Li [18] and Zhang [19] studied a vertical information exchange in a supply chain where demand in-
formation is uncertain to the supplier. Their focus was on the effects of horizontal information leakage for vertical
information sharing in a supply chain, i.e., if one retailer vertically discloses his information to the supplier, the latter
will react to the revealed information, and make decisions accordingly. Li and Zhang discovered that the other retailer
can infer his competitor’s private information through the supplier’s decision.
840 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851

However, the research assumes that retailers are homogeneous, that is, the retailers sell the same product without
differentiating the product through valued-added practices. Because adding value is crucial in the supply chain practice
as illustrated in previous examples, in this paper we do not explicitly consider horizontal information leakage, but
mainly focus on examining the appropriate value to be added to a generic product. We are interested in investigating
the equilibriums in our supply chain structure with value-adding retailers and under what conditions the retailers have
incentives to share their information vertically with the supplier.
The main contributions of our research to the extant literature are threefold: first we find that equilibrium points exist
in a supply chain with value-adding (heterogeneous) retailers under certain conditions; second, we show the upstream
player always benefits from more accurate information about the retailers. Third, we prove information sharing could
benefit both upstream and downstream players, thus leading to a win–win situation for the whole supply chain.

2. Model framework

We consider a supply chain consisting of one supplier and two heterogeneous retailers. We assume there is no
wholesale price discrimination. The supplier sells a common generic product to two retailers at the same wholesale
price. Afterwards, the retailers add their own values to the product and decide on the retail prices to the customers.
Here each retailer has his own private information about the cost of the value added to the product, which is unknown
to the other players. For example, each retailer has his private warranty cost, or each franchisee has his private cost
information of the sale. Considering the retailers are competitors, we assume there is no horizontal information sharing
between retailers; however, each retailer can make his own decision if he is willing to disclose his cost information
to the supplier vertically. Furthermore we assume each party only takes into account the direct effects of information
sharing between different players. As a result, there is no horizontal information leakage considered in this paper. In
fact, in the real situation, one player (i.e., the supplier) makes decision based on many considerations (e.g., forecast,
cost structure, market promotional strategy, brand consideration, etc.). Thus, it is very hard for a retailer to infer the
other retailer’s private information based on the supplier’s decision.
We model our supply chain as a three-stage game. The sequences of moves are as follows:

(1) Each retailer has his own private cost information about value-added service, and decides if he is willing to share
the cost information with the supplier vertically.
(2) Informed of the retailers’ cost information, if any, the supplier announces the wholesale price to optimize his
(expected) profit. In this stage, the supplier may have either, both, or neither retailer’s cost information. Therefore,
three different scenarios exist for the wholesale price.
(3) In response to the wholesale price, two retailers decide on the values added to the product and the retail prices
simultaneously to maximize their own expected profits, respectively.

Next, we describe three key system components in our supply chain framework: demand functions, value-added
functions and payoff functions. Before that, we will summarize our notations as follows:
w: The supplier’s wholesale price to the retailers (decision variable)
p1 : Retailer 1’s retail price to the customers (decision variable)
p2 : Retailer 2’s retail price to the customers (decision variable)
v1 : Value added to the retailer 1’s product (decision variable)
v2 : Value added to the retailer 2’s product (decision variable)
d1 : Demand via retailer 1’s channel
d2 : Demand via retailer 2’s channel

2.1. Demand functions

In this paper, we adopt linear demand functions mainly because linear demand functions are widely used in supply
chain and marketing literature due to their tractability. Specifically we use demand functions similar to those proposed
by Tsay and Agrawal [10]. It is worth noting that this kind of linear model structure has been empirically tested by
other researchers. For example, Raju et al. [20] employed a similar demand structure to study market characteristics
when retailers introduced store brands to compete against national brand products. Sayman et al. [21] used the same
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 841

linear model structure to study the position of store brands against national brands. In essence, the structure of our
demand function is similar to theirs. For other demand functions, if they can be approximated or locally approximated
by linear functions, our managerial insights still hold.
For retailer 1:

d1 = 1 − 1 p1 + 2 v1 + 1 (p2 − p1 ) + 2 (v1 − v2 ). (1)

For retailer 2:

d2 = 2 − 1 p2 + 2 v2 + 1 (p1 − p2 ) + 2 (v2 − v1 ). (2)

i (i = 1, 2) are market bases for two retailers. 1 measures the marginal demand change with channel price. 2
measures the marginal demand change with value-added service. i (i = 1, 2) are migration rates for the perceived price
difference and the perceived service difference, which means if customers perceive there is price difference (value-
added service) between two retailers, they will switch to the other retailer at the rate of 1 (2 ). Therefore, horizontal
competition exists between the two retailers. We also make the following assumptions about the parameters in the
demand functions:
(1) Market base i (i = 1, 2) is large enough compared to other market parameters. Typically this assumption holds
for real life cases.
(2) We assume i > i (i = 1, 2), which means that the effect of a direct price/value change is greater than the effect
of a price/value difference between the two channels (cross-price/value change).

2.2. Value-added functions

In practice, we can use a strictly convex service function c(v) to depict the relationship between the added value and
its related cost. That is, the cost of value-added service have the properties of dc(v)/dv > 0 and d2 c(v)/dv 2 > 0. One
form commonly adopted in previous literature (e.g., [10,22]) is given Eq. (3)

v2
ci (v) = i , i = 1, 2. (3)
2
i is the efficiency parameters of retailer i’s cost structure. The smaller the i , the more efficient retailer i is. In this
paper, we also assume that each retailer knows his own i but does not know the other retailer’s efficiency parameter.
However, each retailer knows the probability distribution of the other retailer’s cost information (i ) (e.g. [23]). The
rationality is as follows: each player has no complete information about a specific retailer’s cost structure; but he
can estimate the probability distribution of the parameter from the whole industry’s information. For simplicity, in
this paper, we assume i is uniformly distributed, i.e., i ∼ U [¯ − ε, ¯ + ε]. Each player can estimate the average
value-added cost efficiency ¯ and the deviation ε. The larger the ε, the more dispersed the industry is (usually in hi-tech
industries). In our model, 1 is known only to retailer 1, 2 is known only to retailer 2, and neither 1 nor 2 is known
to the supplier.

2.3. Profit functions

Without loss of generality, we assume marginal production cost for the supplier is 0. Thus the supplier’s profit can
be expressed as

M = (d1 + d2 )w. (4)

For the retailers, the profit functions are given in Eq. (5)

Ri = [pi − w − ci (v)]di , i = 1, 2. (5)

As we mentioned before, this is a three-stage game. We analyze this dynamic game with the backward induction
approach. First we assume the wholesale price is given and find the equilibrium retail prices and added values in
stage 3.
842 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851

3. Stage 3: retailers’ equilibriums

Because we assume there are no horizontal information sharing and no information leakage, each retailer has no full
information about the other competitor’s cost information. Therefore for each retailer, the objective is to find optimal
retail price pi and the optimal added value vi given the wholesale price w, i.e.

Max E[Ri |w] = Max E[(pi − w − ci )di |w], i = 1, 2. (6)

pi ,vi pi ,vi

Also we assume two retailers decide on the retail prices and added value simultaneously. Therefore, we have the
following proposition.

Proposition 1. There exists equilibrium retail price and added value for each retailer.

Proof. See Appendix. 

The managerial insight of Proposition 1 is that both retailers can have stable retail prices and added values even if
they are heterogeneous, i.e., the retailers should have stable marketing strategies. No one can be better off by deviating
from these pricing and value-added strategies. Also the optimal value added to each retail channel is independent of
the wholesale price w, which is decided only by the retailer’s own cost efficiency (i ). This result will not be affected
even if horizontal information leakage occurs. The more efficient the retailer is, the more value he is willing to add to
the product, which is consistent with intuition.
On the other hand, the retail price is related to the supplier’s wholesale price, the cost efficiency and the dispersion
of the industry’s efficiency. Depending on the different wholesale prices (Stage 2 in next section shows there are three
different wholesale prices for information sharing cases), the retail prices vary accordingly. The retailer tends to charge
more to the customers if the wholesale price is higher, and the retail price is higher if the retailer is more efficient due
to more added values. Furthermore we have the following corollary regarding the retail prices.

Corollary 1.
⎧ jp∗
⎪
⎨
i
>0 if 32 1 − (41 + 1 )2 > 0,
jε
∗
⎩ jpi < 0
⎪
if 32 1 − (41 + 1 )2 < 0.
jε

Proof. See Appendix. 

In order for 32 1 − (41 + 1 )2 > 0, one possibility is that 2 is larger and 1 is smaller, which indicates that
customers are more value-oriented than price-oriented. The managerial insight of Corollary 1 is that the retail prices
will go up if customers are more sensitive to the added value and the industry is more dispersed. In the hi-tech industry
(computer, electronics products), for example, if customers need more added value (after sale service, maintenance,
technical training, etc.), and the industry is highly dispersed (as in the introduction stage of every new technology), the
retail price is usually higher. Afterwards when there is not much efficiency difference between retailers, the retail price
will go down. On the other hand, if customers are more sensitive to the price rather than the added value, the retailers
have lower retail prices if the whole retail industry is highly dispersed.
Next we will find the supplier’s optimal decision given the best response functions obtained above.

4. Stage 2: supplier’ equilibrium

In the second stage, the supplier needs to decide the wholesale price w to maximize his own anticipated profit based
on the information available on hand about the retailers’ cost structure. Eq. (7) is the supplier’s objective function.

Max E[M ] = Max E[(d1 + d2 )w]. (7)

w w
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 843

Because each retailer will decide if he is willing to reveal his private information to the supplier in the first stage, we
have three different cases in the second stage, as follows:

(1) Both of the retailers will share information with the supplier.
(2) One of the retailers will share information with the supplier.
(3) No any retailer will share information with the supplier.

In what follows, we will discuss each of these cases, respectively.

4.1. Case 1: vertical information sharing between the supplier and both retailers

If both retailers decide to share information with the supplier in the first stage, the supplier has full knowledge of the
retailers’ value-added cost information. Hence the supplier’s objective function is

Max M = (d1 + d2 )w. (8)

w

Proposition 2 (The supplier’s equilibrium wholesale price under complete information sharing).

∗ (1 + 2 ) [22 1 + 1 (2 − 2 )](2 + 2 ) [22 1 + 1 (2 − 2 )](2 + 2 )

wCI = + + .
41 81 (1 + 1 )2 1 81 (1 + 1 )2 2

Proof. See Appendix. 

Since we assume 2 > 2 , it is obvious that the wholesale price decreases in i (i = 1, 2). The managerial insight of
Proposition 2 is that if both retailers are willing to disclose their private information to the supplier, or if a manufacturer
like IBM, Sony, etc., has monopoly power in the supply chain, and he can coerce the retailers to share their cost
information with the manufacturer, then the wholesale price is the function of [1/1 + 1/2 ], which indicates that
the more efficiently the retailer can add value to the product, the higher the wholesale price is. In other words, the
manufacturer will be better off by acquiring information from efficient retailers. By contrast, the retailers will be worse
off since they always prefer to have a lower wholesale price.

4.2. Case 2: no vertical information sharing between the supplier and any retailer

Suppose there is no information sharing between the supplier and any of the retailers. The supplier’s objective is to
find the optimal wholesale price w to maximize his expected profit, i.e.,
 ¯ +  ¯ +
Max E[M ] = (d1 + d2 )wf (1 )f (2 ) d1 d2 . (9)
w ¯ − ¯ −

Accordingly, we have Proposition 3.

Proposition 3. The supplier’s optimal wholesale price is given as

¯ + ε
[22 1 + 1 (2 − 2 )](2 + 2 ) ln
∗ (1 + 2 ) ¯ − ε
wNI = + .
41 81 (1 + 1 )2 ε

Proof. See Appendix. 

Proposition 3 provides us the optimal wholesale price if the manufacture has no complete information about the
retailers’ cost structure. The managerial insight of Proposition 3 is that if no retailer is willing to reveal his information
to the upstream supplier, the wholesale price is decided by the average industry’s cost efficiency. Also it can be proved
jwNI∗ /jε > 0, which means that the more dispersed the value-added cost structure of the retailing industry (or more
844 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851

differentiated the retailers are), the higher the wholesale price is. In other words, the supplier could be better off in
more heterogeneous retailing markets.

4.3. Case 3: vertical information sharing between one retailer and the supplier

Without loss of generality, we assume retailer 1 decides to share information with the manufacture due to symmetric
characteristics, i.e., the supplier has full knowledge of the retailer 1’s cost structure 1 . Thus the objective of the supplier
is to find an optimal wholesale price to maximize his expected profit.
 ¯ +ε
Max E[M ] = (d1 + d2 )wf (2 ) d2 . (10)
w ¯ −ε

Proposition 4. If one retailer discloses his cost information with the supplier, the optimal wholesale price is given as

∗ 1 +  2 2(2 + 2 )[32 21 + 21 (2 − 32 ) + 1 1 (52 − 22 )]

wPI = +
41 81 (1 + 1 )2 (21 + 31 )i
(2 + 2 )[32 21 + 21 (2 + 2 ) + 1 1 (22 − 2 )] ¯ + 
+ ln (i = 1, 2).
81 (1 + 1 ) (21 + 31 )
2 ¯ − 

Proof. Same as Proposition 3. 

Also it can be proved that jwPI ∗ /jε > 0, which suggests that the wholesale price increases in the value-added cost

dispersion if only one retailer is willing to disclose his cost information.

5. Equilibrium study in the first stage

In this section, we will study the retailers’ information revelation mechanism in the first stage, i.e., under what
conditions both retailers are willing to share their cost information with the supplier; under what conditions neither
retailer is willing to share his cost information; and under what conditions one of the retailers is willing to share his
cost information while the other is not. Also we will investigate if equilibrium exists in the first stage. First we will
compare the supplier’s profits under different vertical information sharing cases.

5.1. Supplier’s profits

We define the supplier’s profit notations according to the different cases:

M (0): The supplier’s profit under the case of no vertical information sharing with any retailer.
M (1): The supplier’s profit under the case of vertical information sharing with one retailer.
M (2): The supplier’s profit under the case of full vertical information with both retailers.

Proposition 5. ∗M (2) ∗M (1) and ∗M (2)∗M (0)

Proof. See Appendix. 

This proposition indicates that the supplier is always better off with full vertical information sharing than with no
information sharing or with partial information sharing. This result is consistent with other research (e.g. [18,19]).
Furthermore we are interested in comparing the partial information case and the no information sharing case, i.e.,
is the supplier always better off by obtaining partial information than no information at all? Towards this, we have the
following proposition.
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 845

Proposition 6. Suppose retailer i (i = 1, 2) reveals his information to the supplier, and if

2ε
3−i < i < ˜ or 3−i > i > ˜ then ∗M (1) > ∗M (0) where ˜ =  .
¯ + ε
ln
¯ − ε

Proof. See Appendix. 

This proposition suggests that if the supplier shares information with one retailer, for example with retailer 1,
his profit will be better off under the following two situations: (1) retailer 1’s value-added cost is efficient enough
(1 < ˜ ) and the other retailer is more efficient enough; (2) retailer 1’s value-added cost is not efficient enough (1 > ˜ )
and the other retailer is even worse. To sum up, the supplier is better off by accessing one retailer’s information
only if the other retailer has higher (smaller) cost parameter than this retailer when this retailer’s cost is higher
(smaller) than the threshold value ˜ . Therefore even if one retailer is willing to disclose his private information
vertically, the supplier may not accommodate that information into his decision process because the supplier may not be
better off.
Also it can be proved d˜ /dε < 0, that is, the more dispersed the retailing industry is, the smaller the threshold
should be.

5.2. Retailer’s profits

In this section, we will establish conditions under which both retailers have incentives to disclose their private
information, or neither retailer would share information vertically.
Similar to Zhang [19], we first define the following notations for the retailers’ profits:
Ri (1s , 2n ): Retailer i’s (i = 1, 2) profit under the condition that retailer 1 shares information with the supplier while
retailer 2 does not.
Ri (1n , 2n ): Retailer i’s profit under the condition that neither retailer shares information with the supplier.
Ri (1s , 2s ): Retailer i’s profit under the condition that both retailers share information with the supplier.
Ri (1n , 2s ): Retailer i’s profit under the condition that retailer 2 shares information with the supplier while retailer
1 does not.
Correspondingly we have the following conclusions:
(1) For retailer 1: R1 (1s , 2n ) < R1 (1n , 2n ) and R1 (1s , 2s ) < Ri (1n , 2s ),
For retailer 2: R2 (1n , 2s ) < R2 (1n , 2n ) and R2 (1s , 2s ) < R2 (1s , 2n ),
which means that each retailer, if he will be worse off by sharing information with the supplier regardless of the other
retailer’s decision, will not be willing to reveal his private information vertically. As a result, no vertical information
sharing is a unique equilibrium in the first stage of game. In the franchising industry, each franchisee holds private cost
information. The franchisee will receive a higher margin if he has greater advertising effectiveness. Therefore he is not
willing to reveal his private information.
(2) For retailer 1: R1 (1s , 2n ) > R1 (1n , 2n ) and R1 (1s , 2s ) > R1 (1n , 2s ),
For retailer 2: R2 (1n , 2s ) > R2 (1n , 2n ) and R2 (1s , 2s ) > R2 (1s , 2n ),
which means if each retailer will be better off by sharing information with the supplier regardless of the other retailer’s
decision, both retailers are willing to reveal their private information vertically. Therefore vertical information sharing
is a unique equilibrium in the first stage of game. This situation is desirable for the whole supply chain because every
player will be better off by information sharing.
For simplicity of exposition, we assume 1 = 2 = , 1 = 2 =  and 1 = 2 =  to analyze the retailers’ profits.
Accordingly we have the following propositions:

Proposition 7. (1) If the market base  is large enough, 1 < ˜ and 2 < ˜ , no information sharing is the unique
equilibrium in the first stage of the game.

Proof. See Appendix. 

846 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851

The managerial insight of Proposition 7 is that both retailers are unwilling to disclose their cost information if
they are efficient. The intuition underlying the impact of cost structure information sharing on retailers’ profits is as
follows. When retailers are cost efficient, i.e., 1 < ˜ and 2 < ˜ , the supplier sets a higher wholesale price in the
case of information sharing than it does in the case of no information sharing. This is because the wholesale price,
from Proposition 2, is a decreasing function of 1 and 2 . Naturally, retailers would have incentive to share their cost
information so as to motivate a lower wholesale price. However, when 1 < ˜ and 2 < ˜ , the wholesale price in the
case of information sharing is higher than the price in the case of no information sharing, so retailers would have
no interest in revealing their information. This underscores the importance of the threshold value ˜ for the retailers’
decision-making process. Thus these results give us important insights into the channel dynamics under asymmetric
information.

Proposition 8. If the market base  is large enough, 1 > ˜ and 2 > ˜ , vertical information sharing is the unique
equilibrium in the first stage of the game.

Proof. Same as Proposition 7. 

The managerial insight of Proposition 8 is that both retailers have incentive to share cost information, which also
benefits the supplier because the supplier does not need to pay either retailer for his private cost information. At the
same time the retailer’s profit can be improved as well. Therefore this is a win–win situation for all players, and there
is no information asymmetry in this case.

Proposition 9. If the market base  is large enough, 1 > ˜ and 2 < ˜ , or vice versa, then one retailer is willing to
disclose information to the supplier while the other is not.

Proof. Same as Proposition 7. 

Note Proposition 9 is not necessary an equilibrium due to the following reason: suppose 1 > ˜ and 2 < ˜ , then
retailer 1 is willing to share information with the supplier while retailer 2 is not. From Proposition 6, it is possible
that the manufacturer may not accept the retailer 1’s information because he may not be better off by only acquiring
one retailer’s information. The managerial insight of Proposition 9 is that equilibrium may not exist if one retailer
is efficient while the other is not. The supplier’s decision about the wholesale price may be based on both retail-
ers’ or only one retailer’s information. Therefore, it is impossible for one retailer to infer the other’s private cost
information through information leakage. This also can justify our assumption of no information leakage in this
research.
However, since the supplier is always better off by sharing information with two retailers rather than one, one natural
question arises: If only one retailer is willing to reveal his private information to the supplier, can the supplier induce
the other retailer to disclose his information by some side payment? For example, retailer 1 is willing to disclose his
cost information while retailer 2 is not. From Proposition 9, we know that n1 < ˜ and 2 < ˜ . In order for the supplier
to benefit from information sharing by side payment, the following condition must be satisfied:
M (1s , 2s ) − M (1s , 2n ) > R2 (1s , 2n ) − R2 (1s , 2s ).
Here the left side is the upper bound the supplier is willing to pay, and the right side is the loss retailer 2 incurs due to
disclosing his cost information.
Towards this, we have the following Proposition.

Proposition 10. If one retailer is willing to disclose his cost information while the other is unwilling to, the supplier
always can not afford to pay the latter to obtain his private cost information.

Proof. See Appendix. 

The managerial insight of Proposition 10 is that although the supplier is always better off by accessing two retailers’
cost information, he cannot gain enough to cover the loss of the retailer who is unwilling to disclose his information.
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 847

Simply put, side payment is not a strategy to extract more accurate information from a downstream player. Actually, if
a retailer is cost efficient, he can gain more by not sharing information with the supplier. For example, a manufacturer
supplies a certain product (e.g., Haier washer) to two retailers/distributors. If one is a very efficient retailer/distributor,
like Wal-Mart, it will not share its private information with the supplier. Even if the supplier is willing to pay some side
payment to Wal-Mart for its private information, the extra payoff the supplier gains by obtaining Wal-Mart’s private
information can not compensate for Wal-Mart’s loss.

6. Concluding remarks

In this paper, we study a supply chain system with one supplier and two value-adding retailers. Both retailers have
private information about their own cost structure that is unknown to the supplier and the other retailer. We model the
supply chain problem as a three-stage game and prove that under certain conditions, equilibrium exists. We also find
that the supplier is always better off by obtaining private information from both retailers. Furthermore we find that both
retailers may have incentive to share their private information with the supplier under certain conditions, thus leading
to a win–win situation for the whole supply chain.
This research can be extended in several directions. First, because the supplier is better off sharing information with
both retailers, if neither of the retailers reveals their information, we may study if it is feasible for the supplier to use
side payment to pay just one, then which one, or both of the retailers to acquire information. Second in this paper, we
assume cost efficiency is uniformly distributed; we may relax this to other probability distributions and investigate if
the managerial insights obtained here are applicable to other situations. Another limitation of this research is we do
not consider the effect of horizontal information leakage for vertical information sharing in a supply chain [24], i.e.,
each retailer may infer the other’s cost structure through the wholesale price announced by the supplier, which could
be a very interesting topic in the future research. However, if there are numerous heterogeneous retailers in the market,
it is very hard for one retailer to infer the others’ private cost information.

Acknowledgements

We thank anonymous reviewers and the associate editor for their helpful comments on the revision of the paper.

Appendix A

A.1. Proofs of All Results

Proof of Proposition 1. For the retailer 1:

 ¯ +  ¯ +ε
1
Max E[R1 ] = E[(p1 −w−c1 )d1 ] = (p1 −w−c1 )d1 f (2 ) d2 = (p1 −w−c1 ) d1 d2 . (A.1)
p1 ,v1 ¯ − 2ε ¯ −ε
For the retailer 2:
 ¯ +  ¯ +ε
1
Max E[R2 ] = E[(p2 −w−c2 )d2 ] = (p2 −w−c2 )d2 f (1 ) d1 = (p2 −w−c2 ) d2 d1 . (A.2)
p2 ,v2 ¯ − 2ε ¯ −ε
From the first-order condition (FOC), we have the optimal response function for retailer 1:
jE[R1 ] jE[R1 ]
= 0, = 0,
jp1 jv1

1 3( +  ) 2
p1∗ = 21 + 21 w + 21 p2 + 21 w − 22 v2 + 2 2
, (A.3)
41 + 41 (1 + 1 )1

2 + 2
v1∗ = . (A.4)
(1 + 1 )1
848 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851

For retailer 2, the optimal response function is obtained from

   
jE R2 jE R2
= 0, = 0,
jp2 jv2

∗ 1 3(2 + 2 )2
p2 = 22 + 21 w + 21 p1 + 21 w − 22 v1 + , (A.5)
41 + 41 (1 + 1 )2

2 +  2
v2∗ = . (A.6)
(1 + 1 )2
Eq. (A.3) is the best response function of retailer 1. Since retailer 1 has incomplete information of retailer 2, he has no
complete knowledge about retailer 2’s price p2 . Thereby retailer 1 will find the expected retail price for retailer 2.
 ¯ +ε

1 3(2 + 2 )2
p2 = 22 + 21 w + 21 p1 + 21 w − 22 v1 + f (2 ) d2 . (A.7)
¯ −ε 41 + 41 (1 + 1 )2

Similarly the retailer 2 will find the expected retail price for retailer 1.
 ¯ +ε

1 3(2 + 2 )2
p1 = 21 + 21 w + 21 p2 + 21 w − 22 v2 + f (1 ) d1 . (A.8)
¯ −ε 41 + 41 (1 + 1 )1

Substituting (A.7) into (A.3), (A.8) into (A.5), and solving (A.3), (A.5) simultaneously, we obtain the equilibriums for
the retailers:
⎧ A C ¯ + ε
⎪
⎨ p1∗ = + B1 + ln ,
1 ε ¯ − ε
(A.9)
⎩ v ∗ = 2 + 2
⎪
1 (1 + 1 )1
and
⎧ A C ¯ + ε
⎪
⎨ p2∗ = + B2 + ln ,
2 ε ¯ − ε
(A.10)
⎩ v ∗ = 2 + 2 ,
⎪
2 (1 + 1 )2
where
(2 + 2 )[31 (2 + 2 ) + 1 (32 + 22 )]
A= ,
(1 + 1 )(21 + 1 )(21 + 31 )
2 1 + 21 (1 + 1 ) + w(1 + 1 )(21 + 31 )
B1 = ,
(21 + 1 )(21 + 31 )
1 1 + 22 (1 + 1 ) + w(1 + 1 )(21 + 31 )
B2 = ,
(21 + 1 )(21 + 31 )
(2 + 2 )[32 1 − (41 + 1 )2 ]
C= . 
4(1 + 1 )(21 + 1 )(21 + 31 )

Proof of Corollary 1. We only need to prove that

⎡ ⎤
¯ + ε
ln
⎢ ¯ − ε ⎥
d⎢
⎣
⎥
⎦
ε
> 0.
dε
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 849

It can be derived that

⎡ ⎤
¯ + ε
⎢ ln ¯ − ε ⎥
d⎢⎣
⎥
⎦
ε ¯ + ε ¯ + ε
1+ ln
¯ − ε ¯ − ε
= −
dε ε(¯ + ε) ε2
when ε = 0,
¯ + ε ¯ + ε
1+ ln
¯ − ε ¯ − ε
> ,
ε(¯ + ε) ε2
when ε > 0,
⎛ ⎞ ⎛ ⎞
¯ + ε  ¯ + ε 
⎜ 1 + ln
⎜ ¯ − ε ⎟ ⎜
⎟ ⎜ ¯ − ε ⎟
⎟
⎝ ε(¯ + ε) ⎠ > ⎝ ε 2 ⎠ .

Thus we prove
¯ + ε ¯ + ε
1+ ln
¯ − ε ¯ − ε
> ,
ε(¯ + ε) ε2
therefore
⎡ ⎤
¯ + ε
ln
⎢ ¯ − ε ⎥
d⎢
⎣
⎥
⎦
ε
> 0. 
dε
Proof of Proposition 2. Obtained by substituting Eq. (A.9) and Eq. (A.10) into Eq. (8) directly, then set jM /jw = 0
and solve for w because 1 and 2 are both known to the supplier in this case. 

Proof of Proposition 3. Eq. (9) can be simplified as

¯ + ε
w[2(1 + 2 − 21 w)(1 + 1 )2 ε] + w[22 1 + 1 (2 − 2 )](2 + 2 ) ln
¯ − ε
Max E[M ] = . (A.11)
w 2(1 + 1 )(21 + 1 )ε
∗ )
Setting jE[M ]/jw = 0, we obtain the optimal wholesale price under no information sharing case (wNI
¯ + ε
[22 1 + 1 (2 − 2 )](2 + 2 ) ln
∗ (1 + 2 ) ¯ − ε
wNI = + .  (A.12)
41 81 (1 + 1 )2 ε

Proof of Proposition 5. We assume the supplier shares information with retailer 1. By subtracting ∗M (1) from ∗M (2)
and simplifying the expression, we obtain
  
¯ +  2
2 2
[32 1 + 1 (2 − 32 ) + 1 1 (52 − 22 )] (2 + 2 ) 2 − 2 ln
2 2
¯ 
∗M (2) − ∗M (1) = 0.
321 (1 + 1 )3 (21 + 1 )(21 + 31 )2 2 22
We can get the same result if retailer 2 shares information with the supplier.
850 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851

With the same approach, we have

  2
¯ +ε
[32 21 +21 (2 −32 )+1 1 (52 −22 )]2 (2 +2 )2 ε(1 +2 )−1 2 ln
¯ −ε
∗M (2)−∗M (0) = 0. 
81 (1 +1 )3 (21 +1 )(21 +31 )2 ε 2 21 22

Proof of Proposition 6. Without loss of generality, we assume retailer 1 reveals his cost information to the supplier.
By some algebra, we obtain
∗M (1)−∗M (0)
     
¯ +ε ¯ +ε
[32 21 +21 (2 −32 )+1 1 (52 −22 )]2 (2 +2 )2 2ε(21 +2 )−31 2 ln 2ε−1 ln
¯ −ε ¯ −ε
=
321 (1 +1 ) (21 +1 )(21 +31 ) ε 1 2
3 2 2 2 2

if
2ε
2 < 1 <  ,
¯ + ε
ln
¯ − ε
we can verify that
 
¯ + 
2 − 1 ln >0 and
¯ − 

¯ + ε
2(21 + 2 ) − 31 2 ln > 0.
¯ − ε
Thus ∗M (1) − ∗M (0) > 0.
Similarly if
2ε
2 > 1 >  ,
¯ + ε
ln
¯ − ε
then
   
¯ + ε ¯ + 
2ε − 1 ln <0 and 2(21 + 2 ) − 31 2 ln <0
¯ − ε ¯ − 
thus ∗M (1) − ∗M (0) > 0.
We obtain the same results if retailer 2 reveals his information to the supplier.

Proof of Proposition 7. First we consider

R1 (1s , 2n ) − R1 (1n , 2n )
⎧  ⎫⎡  ⎤
⎪ ¯ +ε ⎪ ¯ +ε
⎪
⎨ ⎪
¯ − ε ⎬ ⎢
ln ln
⎢ ¯ − ε ⎥
⎥
−(+)2 −4(+)1 +2 [72 +6(+21 )+(15+81 )] − (+)(3+2)1 2 ⎣ 1 − 1 ⎦
⎪
⎪ 2ε ⎪
⎪ 2ε
⎩ ⎭
= .
16(2+)2 (2+3)2 21 2

We assume the market base  is large enough, which is a very realistic assumption. If
⎛ ⎞  
¯ + ε
ln
⎜ 2ε ⎟ ¯ − ε
1 < ˜ ⎜
⎝ 
˜ =   ⎟ , i.e., 1 − 1 > 0,
¯ +ε ⎠ 2ε
ln
¯ − ε
then R1 (1s , 2n ) < R1 (1n , 2n ).
 D.-Q. Yao et al. / Omega 36 (2008) 838 – 851 851

Similarly for R1 (1s , 2s ) − R1 (1n , 2s ) < 0, 1 < ˜ must hold.
For retailer 2, since retailer 2 and retailer 1 are symmetric in our model setting, so the condition is 2 < ˜ . 

Proof of Proposition 10. Without loss of generality, we assume retailer 1 discloses his information to the supplier
while retailer 2 does not. We need to solve the following inequality:

M (1s , 2s ) − M (1s , 2n ) > R2 (1s , 2n )−R2 (1s , 2s ).

By some algebra,
 
M (1s , 2s ) − M (1s , 2n )] − [R2 (1s , 2n ) − R2 (1s , 2s )

(+)2 −2(32 +9+42 )1 +42 [2 +(3−41 )+2(−31 )]
! !

¯ +  ¯ + 
−3(+)1 2 ln 2−2 ln
¯ −  ¯ − 
= .
64(2 + ) (2 + 3)  1 22
2 2 2

Because we assume market base  is large enough, then for [M (1s , 2s )−M (1s , 2n )]−[R2 (1s , 2n )−R2 (1s , 2s )] > 0,
we solve for 2 and obtain the condition that 2 > ˜ , which contradicts our assumption that retailer 2 is unwilling to share
his cost information with the supplier (2 < ˜ ). Therefore it is proved that M (1s , 2s ) − M (1s , 2n ) < R2 (1s , 2n ) −
R2 (1s , 2s ). 

References

[1] Gommel D. Defining Value-added. Industrial Distribution 2004;93(9):58.

[2] Brooks B. Ten ways to add value and defeat price objections. The American Salesman 2003;November:3–6.
[3] Moltzen E. Systemax PC of choice for AOL bundle. CRN 2004;January (1078):47.
[4] Carbone J. Electronics distribution expect more value add from distributors. Purchasing 2003;October:39–43.
[5] Roberts J. Market trends. CRN 2002;October (1018):92.
[6] Anonymous. Distributors see VA services as key to growth. Purchasing 2005;137(7):40.
[7] Hempelmann B. Optimal franchise contracts with private cost information. International Journal of Industrial Organization 2006;24:449–65.
[8] Berner R. Watch out: Best buy and circuit city. Wal-Mart is attacking their business model by offering lower-cost warranties, Business Week
2005;November 21:46–48.
[9] Choi SC. Price competition in a channel structure with a common retailer. Marketing Science 1991;10(4):271–96.
[10] Tsay AA, Agrawal N. Channel dynamics under price and service competition. Manufacturing & Service Operations Management 2000;2:
372–91.
[11] Lee HL, Whang S. Information sharing in a supply chain. International Journal of Technology Management 2000;20(3/4):373–87.
[12] Li L. Cournot oligopoly with information sharing. Rand Journal of Economics 1985;16(4):521–36.
[13] Gavirneni S. Information flows in a capacitated supply chains with fixed ordering costs. Management Science 2002;48(5):644–51.
[14] Cachon G, Fisher M. Supply chain inventory management and the value of shared information. Management Science 2000;46(8):1032–48.
[15] Agrell PJ, Lindroth R, Norrman A. Risk, information and incentives in telecom supply chains. International Journal of Production Economics
2004;90:1–16.
[16] Corbett CJ, Zhou D, Tang CS. Designing supply contracts: contract type and information asymmetry. Management Science 2004;50:550–9.
[17] Lau AHL, Laum H-S, Zhou Y-W. Considering asymmetrical manufacturing cost information in a two-echelon system that uses price-only
contracts. IIE Transactions 2006;38:253–71.
[18] Li L. Information sharing in a supply chain with horizontal competition. Management Science 2002;48(9):1196–212.
[19] Zhang H. Vertical information exchange in a supply chain with duopoly retailers. Production and Operations Management 2002;11(4):531–46.
[20] Raju J, Sethuraman R, Dhar S. The introduction and performance of store brands. Management Science 1995;41(6):957–78.
[21] Sayman S, Hoch SJ, Raju JS. Position of store brands. Marketing Science 2002;21(4):378.
[22] Iyer G. Coordinating channels under price and nonprice competition. Marketing Science 1998;17:338–55.
[23] Ha AY. Supplier–buyer contracting: asymmetric cost information and cutoff level policy for buyer participation. Naval Research Logistics
2001;48:41–64.
[24] Li L, Zhang H. Confidentially and information sharing in supply chain coordination. Working Paper.