Journal of the Operational Research Society (2010) 61, 1645 --1653 © 2010 Operational Research Society Ltd.

y Ltd. All rights reserved. 0160-5682/10

www.palgrave-journals.com/jors/

Retailer or e-tailer? Strategic pricing and

economic-lot-size decisions in a competitive supply
chain with drop-shipping
WK Chiang1∗ and Y Feng2
1 City University of Hong Kong, Kowloon, Hong Kong; and 2 University of Maryland, MD, USA
Drop-shipping is an arrangement whereby an e-tailer, who does not hold inventories, processes orders and
requests a manufacturer to ship products directly to the end customers. To explore the economic benefits of
adopting drop-shipping distribution strategy in a competitive environment, we investigate the profitability and
the efficiency of the drop-shipping channel as compared to the traditional channel. Specifically, we develop
Economic Order Quantity (EOQ) games with pricing and lot-sizing decisions to examine the strategic inter-
actions between a manufacturer and its retailer/e-tailer in the traditional/drop-shipping distribution channels.
We identify conditions under which the drop-shipping channel profitably outperforms the traditional one. It
is found that the economic interests of adopting drop-shipping distribution for the channel members may not
always be consistent. There are cases where only the manufacture would favour drop-shipping. In this study,
we also reveal that the inefficiency caused by lack of coordination in the traditional channel can be alleviated
in the drop-shipping channel where the lot-sizing decision is made by the manufacturer.
Journal of the Operational Research Society (2010) 61, 1645 – 1653. doi:10.1057/jors.2009.139
Published online 28 October 2009

Keywords: distribution channel; inventory; supply chain analysis; conflict analysis; game theory

1. Introduction Although adopting the drop-shipping distribution can result

The Internet has opened new opportunities for supply chain in a lower inventory-related cost, it may possibly discourage
management, meanwhile posing challenges to the practice some potential demands as customers might find it too incon-
of traditional logistics strategies. Many new business initia- venient to buy from an e-tailer due to, for example, the addi-
tives have emerged to take advantages of the Internet by tional waiting time for product delivery. Moreover, as most
substituting or complementing the traditional channel of supply chains operate as a collection of independent channel
distribution with an innovative logistics strategy called the members whose respective profits are in conjunction with each
drop-shipping distribution. Drop-shipping distribution is an individual firm’s price and/or inventory decisions, the impact
arrangement whereby an online retailer (henceforth, we call of adopting the drop-shipping distribution on the strategic
it an e-tailer for brevity) takes customer orders and requests a interactions among channel members is ambiguous. Thus, it
manufacturer/distributor to ship products directly to the end is not clear immediately whether the gain from a more effec-
customers. Obviously, one distinguishing feature of such a tive inventory control can outweigh the loss caused by vertical
distribution strategy is that an e-tailer, by shifting the inventory channel competition in a supply chain. To enhance our under-
management burden to its manufacturers or suppliers, does standing on the economic values of the drop-shipping strategy,
not hold any inventory. A recent survey indicates that more the objective of this study is to develop analytical models
than 30% of online-only retailers use drop-shipping as the that provide justifications on the circumstances when adopting
primary way to fulfill orders (eRetailing World, 2000). It has drop-shipping distribution that could lead to significant busi-
also been reported that many companies in the information- ness values in a competitive supply chain.
technology hardware industry use drop-shipping to keep costs
down (Fuscaldo, 2003). Could adopting the drop-shipping 2. Related literature
distribution indeed enhance channel profitability and distri-
bution efficiency in a supply chain? If so, is such a logistics Past studies have provided various valuable insights into the
arrangement always desired by all channel members? issues related to the implementation of drop-shipping distribu-
tion. In particular, based on the news-vendor type of inventory
model, Netessine and Rudi (2006) examine the competition
∗ Correspondence: WK Chiang, Department of Management Sciences, City between a traditional channel and a drop-shipping channel
University of Hong Kong, Kowloon, Hong Kong. and show that, in most cases, drop-shipping channel is more
1646 Journal of the Operational Research Society Vol. 61, No. 11

attractive than the traditional retail channel. Zhao and Cao Traditional Channel Drop-Shipping Channel
(2004) investigate the competition between a zero-inventory
e-tailer and a positive-inventory one, and they find that the Manufacturer Manufacturer
former charges lower prices, though the price differential Inventory Inventory
decreases if the market expands rapidly. Based on consumer
heterogeneity, Pan et al (2002) study the channel competi-
Retailer
tion and argue that the traditional retailer may provide better E-tailer
Inventory
service, charge a higher price and earn greater profit than the
pure-play e-tailer. Within a single-period framework, Khouja
(2001) examines the mixed strategy in which e-tailers can use
local inventory as a primary source and use drop-shipping for Customer Customer
backup. Under different sources of uncertainty (eg demand
variability and lead-time variability), Ayanso et al (2006) Product Flow Information Flow
develop a simulation model to provide the implications for
Internet retailers to leverage the drop-shipping fulfillment Figure 1 Channel structure. (a) Traditional channel; (b) Drop-
shipping channel.
choice with an inventory rationing policy.
One of the key issues differentiating this work from the
previous studies reviewed above is the lot-size decision. From independent retailer (traditional channel) or an independent
this point of view, our study is related to the EOQ literature on e-tailer (drop-shipping channel). Like conventional EOQ
joint pricing and production decisions. The EOQ model has models in the literature, we assume that the supply chain faces
been widely studied in single-firm optimization but not in a constant customer demands generated by a non-increasing
competitive environment. Whitin’s paper (1955) is one of the price-dependent function. The retailer holds inventory to ful-
earliest works considering the joint pricing and production fill the customer demands at the retail store (see Figure 1(a)),
decisions in an EOQ framework. Kunreuther and Richard whereas the e-tailer, who takes customer orders and initiates
(1971) investigate the interrelationship between the pricing the delivery request, does not hold any inventory (see Figure
and inventory decisions for a retailer who orders products 1(b)). When the e-tailer is adopted as the sales channel, all
from an outside distributor. Abad (1988) extends the work of inventories are stored at the manufacturer and the product is
Kunreuther and Richard (1971) on considering the case when shipped directly from the manufacturer to the end customer.
the supplier offers all-unit quantity discounts. Lee (1993) The basic notation used in our analysis is defined below:
presents a geometric programming (GP) approach to finding
a profit-maximizing selling price and ordering quantity for l: Manufacturer’s production rate, measured by the amount
a retailer. In the multi-period, discrete-time model, constant manufactured in a unit time.
price through the whole planning horizon has been shown G: Time span of the demand. It may contain multiple periods.
optimal under some conditions (Kunreuther and Schrage, d: The customer demand rate. It can be obtained by d =
1973; Gilbert, 2000; Van den Heuvel and Wagelmans, 2006). D( p)/G, where D( p) is the customer demand over the
With different model assumptions, dynamic prices during time span G. D( p) is a function of the retail price p.
different periods have also been widely studied (Thomas, To assure all customer demand can be filled on time, we
1970; Kim and Lee, 1998; Zhao and Wang, 2002). Deng and assume l > d.
Yano (2006) give a comprehensive review on joint decisions K: Manufacturer’s production setup cost. It is a one-time cost
about price and production quantity. during each production cycle, and it is independent of the
To the best of our knowledge, the economic benefits of production quantity.
drop-shipping distribution in an EOQ framework have not yet S: Retailer’s ordering cost. Ordering cost occurs when the
been explored in the literature. This study aims to fill the gap retailer orders products from the manufacturer. This cost
in the literature by developing stylized models to enhance is constant and is not related to the order quantity.
our understanding of this essential subject. Specifically, we h: Retailer’s inventory holding cost rate. This cost rate is the
develop EOQ games with pricing and lot-sizing decisions to retailer’s cost of holding one unit value of the stock. It is
investigate the strategic interactions between upstream and usually calculated based on the interest rate.
downstream supply chain members in the traditional and drop- H: Manufacturer’s inventory holding cost rate. This rate is
shipping distribution channels. similar to retailer’s inventory holding cost rate.
c: Manufacturer’s unit cost of production. It includes the
3. Model development material purchasing cost, the assembling cost, etc.

Consider a two-echelon supply chain where a contract manu- Note that although the framework of model, which consists
facturer (henceforth, we will call this a manufacturer for of a single manufacturer and a single retailer (e-tailer), is
brevity) distributes a standard product either through an quite basic, it serves as a reasonable approximation for some
 WK Chiang and Y Feng—Retailer or e-tailer? Strategic pricing and economic-lot-size decisions 1647

real business context. For example, many contract manufac- Manufacturer’s Inventory Level
turers sell standard products exclusively through a brand- Q
name retailer (eg, Osim does not produce massage chairs,
but it distributes the products for the manufacturer with its
own brand name). Although an exclusive retailer may operate Cycle
Inventory
multiple retail stores, in practice it may establish a regional time
warehouse to fill the demands from local stores. In such a Q Retailer’s Inventory Level
context, the single manufacturer–retailer setup is considered l
applicable. In the case of drop-shipping, the single e-tailer
Q
assumption is less restrictive since e-tailers do not hold any
inventory. As long as the manufacturer’s standard product is
distributed exclusively through a single e-tailer (or multiple
e-tailers with product differentiations), the applicability of our time
Place Receive Place Receive Place Receive
model is justifiable. Order Order Order Order Order Order

3.1. The traditional channel Figure 2 Inventory levels in the traditional channel.
We start our analysis by formalizing the traditional
manufacture–retailer channel. Following a common approach decides the retail price p and the order quantity Q to maxi-
in the related EOQ literature (eg Whitin, 1955; Pekelman, mize its profit given by
1974; Eliashberg and Steinberg, 1987; Abad, 1988), assume
r ( p, Q) = p(N − p) − w(N − p)
that the product demand in the traditional channel is a linear      
function of the retail price expressed by D( p) = N −  p, revenue purchase cost
which reduces to (N − p) Q
− S− hw . (2)
D( p) = N − p, Q 2 
(1)   
ordering cost holding cost
when the parameter  is normalized to 1 without loss of
generality (the unit of measurement of quantity being arbi- The first-order conditions of (2) with respect to p and Q are
trary). The parameter N is a given constant which reflect ⎧
⎪
⎪ S
the size of the market. Similar to the studies in the supply ⎪
⎨ N − 2p + w + = 0,
chain literature (eg, Monahan, 1984; Lal and Staelin, 1984; Li Q
(3)
et al, 1996), suppose that the manufacturer adopts a lot-for-lot ⎪ S(N − p) hw
⎪
⎪
⎩ − = 0.
policy to fulfill the retailer’s orders and the delivery lead-time Q2 2
is assumed to be negligible or constant without loss of gener-
ality. Past studies generally assume that, with the receipt of an The two equations in (3) characterize the retailer’s best reac-
order from the retailer, the manufacturer produces the required tion to the manufacturer’s wholesale price decision. Subject to
quantity of the product with an infinite production rate, so that (3), the problem for the manufacturer is to choose the whole-
the manufacturer does not hold any inventory as the product is sale price w which maximizes its profit given by
immediately transferred to the retailer. However, this assump- (N − p)
m (w) = w(N − p) − c(N − p) − K
tion is relaxed in our analysis. In particular, we assume that       Q
the manufacturer’s production rate is a fixed constant larger revenue production cost   
setup cost
than the demand rate, and thus the manufacturer also holds Q(N − p)
inventories and incurs the inventory holding cost. Figure 2 − H c. (4)
shows the relationships among the ordering, production and  2Gl 
holding cost
the inventory status of the manufacturer and the retailer.
In the traditional channel, both the manufacturer and the Note that unlike the retailer’s holding cost (which resembles
retailer bear inventory setup/ordering and holding costs. that in the traditional EOQ model), the manufacturer’s holding
Since the channel is uncoordinated, the manufacturer and the cost depends on the total customer demand specified in (1)
retailer are independent decision makers, and each looks at over the time span G. The detailed formulation of the holding
its own profit when making decisions, ignoring the collective cost item in (4) is given below:
impact of their decisions on the channel as a whole. Following
Holding Cost = Average Inventory × Unit Holding Cost
the conventional setting for a dyadic channel, we assume Cycle Inventory No. of Cycle
that the manufacturer is the Stackelberg game leader. Specif-      
ically, anticipating the retailer’s choices, the manufacturer Q 2 /2l × (N − p)/Q
= × H c, (5)
moves first in determining the wholesale price w. Given the G

manufacturer’s decision in w, the retailer, as the follower, Time Span
1648 Journal of the Operational Research Society Vol. 61, No. 11

where the cycle inventory is the shaded area illustrated in Manufacturer’s Inventory Level

Figure 2. Production Production Production

And Usage And Usage And Usage
The equilibrium of the game corresponds to the solution Q
of the manufacturer’s problem, which is a non-linear opti- Usage Only Usage Only
mization problem with non-linear constraints. To solve the Imax

problem, we first specify p and w as functions of Q based Maximum

Inventory
on (3):
⎧
⎪
⎪ 2NS + NQ2 h + SQh
⎪
⎪
time
⎨ p(Q) = , T0 Q T1 T0 T1 T0 T1
2(S + Q 2 h) l
(6)
⎪
⎪ NSQ − S 2
⎪
⎪
⎩ w(Q) = . Figure 3 Inventory level in the drop-shipping channel.
hQ3 + SQ
Plugging (6) into (4), we can then convert the manufacturer’s
where the parameter , called the drop-shipping refusal
profit function into the following single-variable function
factor, represents the relative demand sensitivity to the price
of Q:
as compared to that in the traditional channel (c.f. Chiang


1 NQS − S 2 S et al, 2003 for a similar justification of the demand function).
Maximize (w(Q)) = N− −
2 QS + Q 3 h Q When  = 1, customers are indifferent between the two chan-

 nels. A higher value of  implies a lower convenience level
NQS − S 2 K cHQ
× − − − c . of dropping-shipping to customers. With the same price, the
QS + Q 3 h Q 2Gl demand in the drop-shipping channel is lower [higher] than
(7) that in the traditional channel if  > 1 [ < 1]. When  = 1,
The problem now becomes an unconstrained optimization customers are indifferent between the two channels. Note that
problem. It can be verified that the optimality condition of (7) Similar to the ‘acceptance level’ in Chiang et al (2003), the
is a six-degree polynomial equation of Q with the following ‘refusal factor’ in this study is used to distinguish customers’
6 × 6 companion matrix:
⎡ ⎤
2S 2 Gl(cS/N − 2S − K )/Hch2
⎢1 4S 2 (N 2 lG − 2SGlh + SHc/2 − NGlc − KGlh)/NHch2 ⎥
⎢ ⎥
⎢ ⎥
⎢ 1 3S (4Glh − H c)/Hch
2 2 ⎥
⎢ ⎥
⎢ ⎥ (8)
⎢ 1 2SG(HcS/G − 2Klh − 2lN 2
− 2cNl)/NHch ⎥
⎢ ⎥
⎢ ⎥
⎣ 1 2(SGl/N + 2SH/ h − KGl/c)H ⎦
1
The six roots of the polynomial equation, which are the
local maximizers for the optimization problem, are the eigen- attitudes towards the traditional (brick and mortar) retailer
values of the companion matrix in (8). Since there are at most and the e-tailer. The rationale behind is that consumers’
six local rational maximizers, the global optimal solution can willingness-to-pay price will be affected by various channel
be easily identified after the local maximizers are found by attributes, which include product found is in stock, physical
any eigenvalue algorithm. examination of products, immediate possession of products,
To understand the efficiency loss in the decentralized tradi- uncertainty about getting the right item, accepts all forms of
tional channel, we also analyse the performance of the central- payment, helpfulness of salespeople, post-purchase service,
ized traditional channel. The analysis is analogous, and thus exchange-refund policy for returns, ability to compare prod-
the details are relegated to Appendix A for the interest of ucts, speed of selection and purchase, charges for shipping
space. and handling, etc. Therefore, even if the e-tailer (who does not
hold any inventory for immediate possession by customers)
3.2. The drop-shipping channel can display the product to customers in a physical space,
To analyse the performance of the drop-shipping channel, customers’ attitudes towards the two shopping channels may
assume that the e-tailer faces the constant demand gener- not be identical.
ated by an analogous non-increasing price-dependent func- When the drop-shipping channel strategy is adopted, inven-
tion. Specifically, the demand function for the drop-shipping tories are held by the manufacturer only. The e-tailer receives
channel is defined as the orders from the customers and requests the manufacturer
to ship the product directly to the customers. Figure 3 shows
D( p) = N −  p, (9) the production and inventory status of the manufacturer. At the
 WK Chiang and Y Feng—Retailer or e-tailer? Strategic pricing and economic-lot-size decisions 1649

beginning of each production cycle T0 , the manufacturer sets Substituting (13) into (12) results in the following single-
the targeted production quantity to be run size Q. During the variable profit function of Q:
production and usage period (from T0 to T1 ), the manufac-

N − w∗ (Q) K
turer produces and delivers the product to customers. Note m (Q) = w∗ (Q) − c −
that if the demand rate were zero, the inventory would accu- 2 Q
mulate at a rate as shown by the dash line. However, due to (2Gl − N + w∗ (Q))Q
− H c. (15)
the positive demand rate, the actual inventory increase rate, as 4Gl
illustrated by the bold solid line, is lower than that with a zero
It can be verified that (15) is convex-concave, and thus we
demand rate. Therefore, at the end of each production cycle
develop below a similar line-search algorithm proposed by
T1 , the maximum inventory is smaller than the run size Q. In
Abad (1988) to obtain the global optimal solution.
the usage only period (T1 to T0 ), the manufacturer consumes
the remaining inventory to fulfill the customers’ orders. Step 1: Let k = 0 and Q 0 = ∞.
Again, to obtain the equilibrium result in the decentralized Step 2: Compute w∗ (Q k ) using Equation (13).
channel, we start with solving the retailer’s problem. Subse- Step 3: Compute Q k+1 using Equation (14).
quently, we solve the manufacturer’s problem, taking into Step 4: If |Q k+1 − Q k | < , stop. Otherwise let k = k + 1
account the reaction function of the retailer. The manufac- and go to Step 2.
turer, as the game leader, decides the wholesale price w and
the production quantity Q in the first stage of the game. Given The analysis of the centralized drop-shipping channel is
the manufacturer’s decisions, the e-tailer sets the retail price detailed in Appendix B.
p to minimize its profit given by
4. Numerical experiments
r ( p) = ( p − w)(N −  p). (10)
Based on the models developed, we conduct numerical exper-
It is straightforward to verify that the optimal retail price is iments to gain more insights into the difference between the
traditional and drop-shipping channels. To better generalize
N + w
p= . (11) the results, we study 2400 cases formed by the combina-
2 tions of the following parametric values: h = H ∈ {0.02,
Anticipating the e-tailer’s best price response in (11), the 0.04, 0.06, 0.08, 0.1}, G ∈ {30, 40, 50, 60}, S ∈ {10 000,
manufacturer, by choosing w and Q, maximizes its profit spec- 13 000, 16 000, 19 000}, K ∈{30 000, 60 000, 90 000, 120 000,
ified by 150 000, 180 000}, l ∈ 60, 90, 120, 150, 180}. The study first
focuses on the case when the two alternative distribution chan-
m (w, Q) = (N −  p)w − (N −  p)c
      nels are equally convenient to customers, that is, the drop-
revenue production cost shipping refusal factor  = 1. Table 1 summarizes the results
(N −  p) Imax of equilibrium decisions and the subsequent profits for each
− K− Hc member.
Q 2
      The results indicate that the discrepancy of retail prices
setup cost holding cost
between the two channels is not very considerable, but the


N − w K
= w−c−
2 Q Table 1 Equilibrium decisions and profits
(2Gl − N + w)Q
− H c, (12) Minimum Maximum Average
4Gl
Traditional channel
where Imax is the maximum inventory illustrated in Figure 3. Wholesale price 1610 1657 1633
Specifically, Ordering quantity 281 886 494

 Retail price 2326 2346 2332
Q Q N − p (2Gl − N +w)Q
Imax = (l − d) = l− = . Profit: Retailer 6724 14 752 10 094
l l G 2Gl Profit: Manufacturer 7790 29 122 17 427
Profit: Total 14 588 43 780 27 521
Based on (12), the first-order conditions of the manufacturer’s
optimization problem are Drop-shipping channel
Wholesale price 1652 1679 1663


1 K N H Qc Production quantity 1194 8028 3291
w∗ (Q) = + +c− , (13) Retail price 2326 2340 2332
2 Q  2Gl
 Profit: Retailer 7272 15 146 10 617
2Gl K (N − w) Profit: Manufacturer 13 817 29 964 20 622
Q∗ = . (14) Profit: Total 21 088 45 111 31 239
H c(2Gl − N + w)
1650 Journal of the Operational Research Society Vol. 61, No. 11

Profit 1.3
Both Members m
1.2 Prefer Traditional
Manufacturer (Manufacturer’s
25000 Channel
Channel
(Drop-Shipping) Drop-Shipping Conflict Zone Indifferent Line)
1.1
Refusal Factor
20000 r

1.0 Both Members (Retailer’s Channel

Prefer Drop Shipping Indifferent Line)
Manufacturer (Traditional)
Channel
15000 0.9
E-tailer 2 4 6 8 10 12
Setup cost / Ordering Cost
10000 Retailer K/S

Figure 5 Impact of setup/ordering cost on channel choice.

5000
0.9 0.95 1.0 1.05 1.1 1.15 1.2
r Conflict Zone m drop-shipping distribution for the channel members is
Drop-shipping Refusal Factor
conflicting.

Figure 4 Impact of drop-shipping refusal factor on channel 5. Sensitivity analysis of channel choice
profitability.
To generate more insights into how the interplays of various
parameters in the model affect the channel preferences for
average wholesale price in the drop-shipping channel is the manufacturer and the retailer, in this section, we conduct
significantly higher than that in the traditional channel. sensitivity analyses to illustrate the impacts of setup/ordering
Although the e-tailer is charged a higher wholesale price cost and inventory holding cost rate on the threshold values
than the retailer, with the advantage of holding no inventory, of adopting the drop-shipping distribution. Note that unless
the average profit of the e-tailer is 5.18% higher than that in otherwise noted, the same parametric values in section 3 are
the traditional retailer. We also find that the average profit used for the analysis.
of the manufacturer in the drop-shipping channel is 18.33%
higher than that in the traditional channel. This is mainly 5.1. Effect of setup/ordering cost
because that the manufacturer in the drop-shipping channel
enjoys the advantage of controlling the production quantity. In this part of the analysis, we identify the threshold values
We conclude that both channel members are better off with of adopting the drop-shipping distribution with respect to
the drop-shipping strategy when the customers are indif- different ratios of the setup cost to the ordering cost (K /S
ferent between the drop-shipping and the traditional channels ratio). The result, illustrated in Figure 5, indicates that while
( = 1). the K/S ratio does not appear to significantly affect the
Intuitively, the gains from adopting the drop-shipping retailer’s channel preference, the manufacturer’s threshold
strategy will be offset when the drop-shipping refusal factor values of adopting the drop-shipping distribution m increases
 is too high. Now we investigate the impact of the drop- with the K/S ratio. It implies that the manufacturer could
shipping refusal factor  on channel performance. Different benefit more from adopting drop-shipping distribution when
values of , ranging from 0.9 to 1.2 with step value 0.05, the setup cost is high. This is to be expected since the
are used in the analysis. Figure 4 illustrates the profits for inventory-related cost reduction is more considerable with
the retailers and the manufacturers in the two different distri- a higher unit setup cost K in the drop-shipping channel in
bution channels. Not surprisingly, the results show that the which the manufacture has full control over the lot-size
corresponding profits for the manufacturer and the e-tailer decision.
both decrease with the drop-shipping refusal factor  in In Figure 5, the manufacturer’s and the retailer’s channel
the drop-shipping channel. Note that, in Figure 4, m and indifferent lines divide the K /S −  plane into three regions.
r represent the average threshold values of adopting the The shaded area depicts the conflict zone where the manufac-
drop-shipping distribution for the manufacturer and the turer prefers the drop-shipping channel whereas the retailer,
e-tailer, respectively. When  > m [ > r ], the manufacture on the contrary, prefers the traditional channel. Obviously, the
[e-tailer] would prefer the traditional distribution strategy likelihood for the manufacturer to favour the drop-shipping
as the drop-shipping distribution is too inconvenient for distribution is higher.
customers. Our analysis indicates that 1 < r < m , and thus
5.2. Effect of holding cost rate
both channel members are better off with drop-shipping
distribution if  < r . When 12 the customers’ drop-shipping Recall that one of the main differences between the traditional
refusal factor falls between r and m (identified as the channel and the drop-shipping channel is that the e-tailer in
conflict zone in Figure 4), the economic interests of adopting the drop-shipping channel does not hold any inventory. When
 WK Chiang and Y Feng—Retailer or e-tailer? Strategic pricing and economic-lot-size decisions 1651

1.4 Table 2 Competition penalty

1.3 Both Members m Centralized Decentralized Competition
Prefer Traditional (Manufacturer’s
Channel Channel
penalty
1.2
Drop-Shipping Indifferent Line)
Refusal Factor Conflict Zone
Traditional channel
1.1
Ordering quantity 5658 494
1.0 Both Members
r
Retail price 1667 2332
Prefer Drop Shipping (Retailer’s Channel
Indifferent Line) Channel profit 41 146 27 521 33.25%
Channel
0.9
0.02 0.04 0.06 0.08 0.1 0.12 Drop-shipping channel
Inventory Holding Cost Rate (h = H) Production quantity 5308 3291
Retail price 1655 2332
Figure 6 Impact of inventory holding cost rate on channel choice. Channel profit 42 002 31 239 25.62%

would it be more profitable for the channel members to have Competition Penalty
the manufacturer carry all the burden of holding inventory?
26.0%
Now we answer this question by examining the impact of
inventory holding cost rate on the channel preference for the 25.8%
manufacturer and the retailer.
Figure 6 illustrates the channel preferences for the manu- 25.6%
facturer and the retailer under various inventory holding cost
rates. Similar to the effect of the K/S ratio, we find that 25.4%
the manufacturer’s threshold values of adopting the drop- 25.2%
shipping distribution m are consistently higher than the
retailer’s threshold values r . A higher inventory holding 25.0%
cost rate generally corresponds to higher values of m and 0.9 0.95 1.0 1.05 1.1 1.15 1.2
r , though the impact of the inventory holding cost rate on Drop-shipping Refusal Factor
r is relatively insignificant. Again, the result indicates that
Figure 7 Impact of drop-shipping refusal factor on competition
likelihood for the manufacturer to favour the drop-shipping penalty.
distribution is higher. The manufacturer’s and the retailer’s
channel indifferent lines divide Figure 6 into three regions
and the conflict zone where only the manufacturer prefers the two channels ( = 1). In other words, the inefficiency
the drop-shipping channel is identified in the shaded area. caused by vertical channel competition in the traditional
channel is alleviated in the drop-shipping channel where the
6. Analysis of channel efficiency manufacturer takes full control over the lot-sizing decision.
Our further analysis indicates that the competition penalty of
The overall channel profit in a decentralized supply chain, the drop-shipping channel increases with  (see Figure 7),
due to the competitive decision-making process, is typically though the increase rate is not very substantial.
lower than that in a centralized supply chain where the system
performs at the optimal level. To measure the channel effi-
7. Concluding remarks
ciency of the two decentralized channels proposed in this
study, we define the competition penalty as the difference in The objective of this study is to explore the economic bene-
the overall supply chain profits between a decentralized solu- fits of adopting drop-shipping distribution in a competitive
tion and the centralized (system optimal) solution, measured environment. We develop EOQ games with joint pricing
as a percentage of the optimal profit. and lot-sizing decisions to investigate the strategic interac-
With same parametric values specified above, Table 2 tions between a manufacturer and its retailer/e-tailer in the
reports the competition penalty for the traditional channel traditional/drop-shipping distribution channels under various
and the drop-shipping channel. We find that the total channel scenarios. From the perspective of each channel member,
profit of the decentralized traditional channel is 33.25% lower we identify the conditions under which the drop-shipping
than the centralized traditional channel. On the other hand, channel outperforms the traditional channel in terms of
the total channel profit of the decentralized drop-shipping profitability.
channel is, on average, 26% lower than the centralized drop- Different from that in the traditional channel where the
shipping channel. Apparently, the significant discrepancy lot-sizing decision is made by the retailer, the manufacturer
in the competition penalty implies that the drop-shipping in the drop-shipping channel takes full control over the lot-
channel is relatively more efficient than the traditional sizing decision as the e-tailer does not hold inventory. The
channel given that the customers are indifferent between result of our analysis indicates that although both channel
1652 Journal of the Operational Research Society Vol. 61, No. 11

members could benefit from drop-shipping distribution Eliashberg J and Steinberg R (1987). Marketing-production decisions
under some conditions, the manufacturer has a higher likeli- in an industrial channel of distribution. Mngt Sci 33: 981–1000.
hood to favour the drop-shipping distribution. Additionally, eRetailing World (2000). The state of eRetailing (Supplement, March).
comparison of the two alternative channels in terms of the Fuscaldo D (2003). E-commerce: B-to-B—Looking big: How can
online retailers carry so many products? The secret is ‘drop-
difference in the profits between a competitive (decentral- shipping’. Wall Street Journal (Apr): R7.
ized) solution and the system optimal (centralized) solution Gilbert SM (2000). Coordination of pricing and multiple-period
reveals that the drop-shipping channel is more efficient than production across multiple constant priced goods. Mngt Sci 46(12):
the traditional one. This implies that the inefficiency caused 1602–1616.
by lack of coordination in the traditional channel can be alle- Khouja M (2001). The evaluation of drop shipping option for
viated by adopting drop-shipping distribution. Another impli- e-commerce retailers. Comput Ind Eng 41: 109–126.
cation from our result is that when designing a contract to Kim D and Lee WJ (1998). Optimal joint pricing and lot sizing with
fixed and variable capacity. Eur J Opl Res 109(1): 212–227.
coordinate a channel with the drop-shipping distribution,
Kunreuther H and Richard JF (1971). Optimal pricing and inventory
the manufacturer has more bargaining power in division of decisions for non-seasonal items. Econometrica 39(1): 173–175.
cooperative profit (especially when the scenario occurs in the Kunreuther H and Schrage L (1973). Joint pricing and inventory
conflict zone). decisions for constant priced items. Mngt Sci 19(7): 732–738.
Although the game-theoretical model developed in this Lal R and Staelin R (1984). An approach for developing an optimal
paper can be viewed as the primitive prototype in examining discount pricing policy. Mngt Sci 30: 1524–1539.
the pricing and lot-sizing decisions with the drop-shipping Lee WJ (1993). Determining order quantity and selling price by
distribution, we recognize that it is limited in many respects. geometric programming: Optimal solution, bounds, and sensitivity.
Decision Sci 24(1): 76–87.
Future research topics extending this study are possible in
Li SX, Huang ZM and Ashley A (1996). Improving buyer-seller
various ways. For example, we consider only one manufacture system cooperation through inventory control. Int J Prod Econ
and one retailer/e-tailer in our model. Although the stylized 43: 37–46.
model is applicable in some situations and the result is justifi- Monahan JP (1984). A quantity discount pricing model to increase
able for the insight-oriented investigation, it should be useful vendor profits. Mngt Sci 30: 720–726.
to extend the analysis by exploring different channel struc- Netessine S and Rudi N (2006). Supply chain choice on the internet.
tures. Another restriction of our model is that it only considers Mngt Sci 52(6): 844–864.
single-product situations. Although such a setting provides a Pan X, Ratchford BT and Shankar V (2002). Price competition
between pure play versus bricks-and-clicks e-tailers: Analytical
starting point for investigating the problem, studies seeking model and empirical analysis. In: Baye MR (ed). Advances in
to tackle multi-product situations will be warranted. In this Applied Microeconomics, Vol. 11, Emerald Group Publishing
study, the demand is modelled as a function of the retail price Limited, pp 29–61.
and is deterministic. It should be informative to incorporate Pekelman D (1974). Simultaneous price-production decisions.
other variables, such as the service level or sales effort in to the Oper Res 22: 788–794.
demand function. The investigation of the impact of demand Thomas J (1970). Price-production decisions with deterministic
demand. Mngt Sci 16(11): 747–750.
variability would also be of interest. Finally, because of the
Van den Heuvel W and Wagelmans APM (2006). A polynomial time
change in scale economies, product transportation costs could algorithm for a deterministic joint pricing and inventory model.
vary with different distribution arrangements. We ignore such Eur J Opl Res 170(11): 463–480.
costs due to the complex nature of the problem. Incorporating Whitin TM (1955). Inventory control and price theory. Mngt Sci 2(1):
relevant transportation costs in to the analysis would certainly 61–68.
be valuable. Zhao H and Cao Y (2004). The role of e-tailer inventory policy on
e-tailer pricing and profitability. J Retailing 80(3): 207–219.
Zhao W and Wang Y (2002). Coordination of joint pricing-production
Acknowledgements — The authors thank the anonymous referees for their decisions in a supply chain. IIE Transactions 34: 701–715.
valuable comments and constructive suggestions.

Appendix A. The centralized traditional channel

References
The profit function of the centralized traditional channel is
Abad PL (1988). Determining optimal selling price and lot size when given by
the supplier offers all-unit quantity discounts. Decision Sci 19(3):
622–634.
Ayanso A, Diaby M and Nair KS (2006). Inventory rationing via N−p
( p, Q) = p(N − p) − c(N − p) − S
drop-shipping in Internet retailing: A sensitivity analysis. Eur J       Q
Opl Res 171: 135–152. revenue production cost   
ordering cost
Chiang WK, Chhajed D and Hess JD (2003). Direct marketing,
indirect profits: A strategic analysis of dual-channel supply-chain N − p H Qc(N − p) Qhc
design. Mngt Sci 49(1): 1–20. −K − −
Q 2Gl 2
Deng S and Yano CA (2006). Joint production and pricing decisions     
with setup costs and capacity constraints. Mngt Sci 52(5): 741–756. setup cost holding cost
 WK Chiang and Y Feng—Retailer or e-tailer? Strategic pricing and economic-lot-size decisions 1653


S K Appendix B. The centralized drop-shipping channel
= p− − − c (N − p)
Q Q The profit function of the centralized drop-shipping channel
H Qc(N − p) Qhc is given by
− − . (A.1)
2Gl 2 ( p, Q) = (N −  p) − (N −  p)c
     
The first-order condition with respect to p, Q are: revenue production cost
⎧ * K +S H Qc
⎪ N − p H Qc(Gl − N +  p)
⎨ * p = N − 2 p + c + Q + 2Gl ,
⎪ − K− .
Q 2Gl
⎪
(A.2)      
⎩ * = (K + S)(N − p) − H c(N − p) − hc .
⎪ setup cost holding cost
*Q Q2 2Gl 2 (B.1)
Then we have the following relationship between the retail The first-order condition with respect to p, Q are:
price and the ordering quantity: ⎧ *( p, Q)
⎧ K H Q c
N +c K +S H Qc ⎪
⎪ = N − 2  p + c + − ,
⎪
⎪ p ∗ (Q) = + + , ⎨ *p Q 2Gl
⎪
⎨ 2 2Q 4Gl (B.2)
 ⎪
⎪
(A.3) ⎩ *( p, Q) = K (N −  p) − H (Gl − N +  p)c .
⎪
⎪
⎪ 2Gl(K + S)(N − p)
⎩Q = ∗
. *Q Q2 2Gl
(hGl + H p − H N )c
Similar to the previous analysis, we have the following rela-
The optimal solutions of p, Q are inter-related. Then we have: tionship between the retail price and the production plan:
⎧

N − p ∗ (Q) H Q c K
1 (Q) = ( p∗ (Q) − c)(N − p∗ (Q)) − (K + S) ⎨ p = N + c − + (2),
Q 2Gl Q (B.3)
⎩
H Qc(N − p ∗ (Q)) Qhc H c(Gl − N +  p)Q 2 = 2Gl K (N −  p).
− − . (A.4)
2Gl 2 Then we get:
The profit function 1 (Q) is a convex-concave function of
N −  p∗ (Q)
Q. A line-search algorithm specified below can be applied to 1 (Q) = ( p∗ (Q) − c)(N −  p∗ (Q)) − K
find the optimal solution for the problem: Q

Step 1: Let k = 0 and Q 0 = ∞. H Qc(Gl − N +  p∗ (Q))

− . (B.4)
N +c K +S H Qk c 2Gl
Step 2: Compute p∗ (Q k ) = + + ,
2 2Q k 4Gl It can be verified that this objective function is also a convex-

2Gl(K + S)(N − p ∗ (Q k )) concave function. A similar line-search algorithm in Appendix
Step 3: Compute Q k+1 = . A can be applied to find the optimal solution.
(hGl + H p ∗ (Q k ) − H N )c
Step 4: If |Q k+1 − Q k | < , stop. Otherwise let k = k + 1 Received April 2008;
and go to Step 2. accepted August 2009 after one revision
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.