PREYAS S.

DESAI, ODED KOENIGSBERG, and DEVAVRAT PUROHIT*

Conventional wisdom in marketing holds that (1) retailer forward buying
is a consequence of manufacturer trade promotions and (2) stockpiling
units helps the retailer but hurts the manufacturer. This article provides
a deeper understanding of forward buying by analyzing it within the con-
text of manufacturer trade promotions, competition, and demand uncer-
tainty. The authors nd that regardless of whether the manufacturer offers
a trade promotion, allowing the retailer to forward buy and hold inven-
tory for the future can, under certain conditions, be benecial for both
parties. Disallowing forward buying by the retailer may lead the manu-
facturer to lower merchandising requirements and change the depth of
the promotion. In competitive environments, there are situations in which
retailers engage in forward buying because of competitive pressures in a
prisoners-dilemma situation. Finally, when the authors consider the case
of uncertain demand, they nd further evidence of strategic forward buy-
ing. In particular, the authors nd cases in which the retailer orders a
quantity that is higher than what it expects to sell in even the most opti-
mistic demand scenario.
Keywords: marketing channels, game theory, trade promotions, pricing,
inventory
ForwardBuyingby Retailers
Forward buying occurs when retailers purchase units dur-
ing a particular period, hold some of them in inventory, and
then sell them in subsequent periods. Because retailer for-
ward buying tends to be correlated with trade promotions
(temporary wholesale price reductions) offered by manu-
facturers, conventional wisdom in marketing suggests that
were it not for these trade promotions, manufacturers would
not need to deal with the problem of retailer forward buying
(e.g., Coughlan et al. 2006; Kotler and Keller 2006). As a
result, researchers have focused their efforts on explaining
why trade promotions are observed and not on why for-
ward buying is not. An exception is Anand, Anupindi, and
Basook (2008), who examine the role of forward buying
in a monopoly setting in which the manufacturer does not
offer a trade promotion. In contrast, we address the effects
*
Preyas S. Desai is Professor of Business Administration (e-mail:
[email protected]), and Devavrat Purohit is Professor of Business Admin-
istration (e-mail: [email protected]), Fuqua School of Business, Duke
University. Oded Koenigsberg is Barbara and Meyer Feldberg Associate
Professor of Business, Columbia Business School, Columbia University
(e-mail: [email protected]). The authors are grateful to seminar par-
ticipants at Carnegie Mellon University, University of Chicago, Dartmouth
College, Harvard Business School, University of California, Los Ange-
les, University of California, Davis, and University of Texas, Dallas. They
are especially grateful to Abel Jeuland for his thoughtful comments. In
addition, they thank the three anonymous JMR reviewers for their helpful
suggestions. Brian Ratchford served as guest editor for this article.
of forward buying on the retailers and manufacturers in a
variety of settings, including competitive environments and
under conditions of demand uncertainty. We also examine
the consequences of preventing retailers from forward buy-
ing during trade promotions.
Forward buying is an important phenomenon in both
marketing and operations literature. In marketing, the gen-
eral view is that because trade promotions are temporary
price discounts, retailers simply stock up on good deals
when they are offered (e.g., Blattberg and Neslin 1990).
The question here is why the price reduction must be tem-
porary and not permanent. Several explanations for this
have been put forth, all of them relying on the competi-
tion among rms. From a theoretical standpoint, the con-
sensus views are that trade promotions persist because of
the intense competition for the switching segment of the
consumer market (e.g., Narasimhan 1988; Raju, Srinivasan,
and Lal 1990; Rao 1990) or because of implicit collu-
sion among manufacturers (e.g., Lal 1990). Note that the
aforementioned models do not include a strategic retailer
in the analysis, so they cannot speak directly to the issue of
forward buying. However, they show that the competition
leads to trade promotions, which tends to be correlated with
forward buying. In contrast, Lal, Little, and Villas Boas
(1996) explicitly model two competing manufacturers that
sell through a common retailer that has an option to for-
ward buy. From a manufacturers perspective, being in the
2010, American Marketing Association
ISSN: 0022-2437 (print), 1547-7193 (electronic)
90
Journal of Marketing Research
Vol. XLVII (February 2010), 90102
Reprinted with permission from the Journal of Marketing Research, published by the American Marketing
Association, Preyas Desai, Oded Koenigsberg, and Devavrat Purohit, volume 47, no. 1 (2010): 90-102.
Forward Buying by Retailers 91
retailers inventory is important because it makes the man-
ufacturers product more competitive for switchers in the
subsequent period. Indeed, it is the presence of the switch-
ing segment that leads to trade promotions and forward
buying. In the absence of switchers, there would be a single
price and no forward buying. More recently, Cui, Raju, and
Zhang (2008) have shown that manufacturer trade promo-
tions can also be a mechanism to price discriminate among
retailers that differ in their holding costs.
Forward buying is also a rich area of study in opera-
tions management, in which researchers have studied rms
inventory decisions in a variety of models. In these studies,
inventory emerges as a mechanism to deal with demand or
supply uncertainty or as a trade-off between ordering and
holding costs. However, the role of inventory as a strate-
gic choice has not been a prominent issue in this literature.
A notable exception is the work of Anand, Anupindi, and
Bassok (2008), who show that inventory plays an impor-
tant strategic role such that a retailer would hold inventory
even when there is no uncertainty about demand. The focus
of their work is on coordinating the supply chain by using
contracts that allow the manufacturer to commit to whole-
sale prices over time. In contrast, we consider the situation
in which manufacturers cannot make credible commitments
not to lower prices in the future. This is the case in most
packaged goods markets; it is possible for manufacturers
to commit not to raise prices, but it is much more difcult
to commit not to lower prices. Furthermore, we examine
more complex competitive channel structures and focus on
the marketing variables of merchandising support and trade
promotions and also on the role of demand uncertainty.
We rst develop a simple model that specically
excludes the standard operations reasons advanced by
researchers for why retailers would forward buy and hold
inventory. Thus, we assume a market in which there is no
uncertainty about demand or supply, no production lead
time, and no ordering or setup costs. Within this frame-
work, we consider both the case when the manufacturer
offers a trade promotion and the case when it does not offer
a trade promotion.
1
From the terms offered by the manu-
facturer, the retailer chooses the quantity to order, the retail
price, and the inventory level. We analyze forward buying
with three channel structures: (1) A single manufacturer
sells to a single retailer; (2) two competing manufactur-
ers sell through a single, common retailer; and (3) a single
manufacturer sells to two competing retailers.
In the single manufacturersingle retailer case, we nd
conditions under which both the retailer and the man-
ufacturer are better-off with forward buying and condi-
tions under which forward buying is protable for the
retailer but not for the manufacturer. In the competitive
case, we allow two manufacturers to sell through a common
retailer and nd that, compared with the previous bilat-
eral monopoly case, forward buying becomes even more
likely. Importantly, each manufacturer reduces wholesale
price in response to the retailers forward buying not only
of its own product but also of the competing manufac-
turers product. As a result, forward buying becomes even
1
As we discuss in greater detail subsequently, a trade promotion in our
model is dened as a temporary price reduction that is contingent on a
specic level of merchandising support provided by the retailer.
more attractive for the retailer. When we introduce compe-
tition at the retailer level and allow a single manufacturer to
sell through two competing retailers, we still nd the pres-
ence of forward buying. We also nd that the competition
between retailers can lead to a prisoners-dilemma situation
such that both retailers are worse-off with forward buying.
This occurs partly because competition forces each retailer
to pass through a greater part of any reduction in whole-
sale price. Furthermore, as the retailer competition becomes
more intense, we nd that the incidence of forward buying
goes down.
Although the bulk of this research rules out other reasons
for forward buying, we note that two of the most common
reasons cited for forward buying are the presence of retailer
trade promotions and uncertainty about demand. Therefore,
we extend our basic framework to specically allow for
trade promotions and the presence of demand uncertainty.
Importantly, we verify that the effects of forward buying
identied previously continue to hold even when the man-
ufacturer offers a trade promotion. Furthermore, we show
that if the terms of the trade promotion prohibit the retailer
from forward buying, the manufacturer will be forced to
require a lower merchandising effort and may also need
to adjust its trade promotion discount. When it comes to
demand uncertainty, forward buying continues to play an
important strategic role. When demand can be either high
or low with specic probabilities, under certain cases, the
retailer orders enough units such that it carries inventory
regardless of the demand state that may arise. This is note-
worthy because conventional wisdom argues that retailers
will end up carrying inventory only when demand turns
out to be low. These two extensions further enhance our
understanding of forward buying.
Finally, most of the channels research in marketing
examines a manufacturerretailer framework in a static set-
ting and one in which the retailers ordering quantity is also
its selling quantity. However, this static setting becomes
inappropriate when retailers can stockpile for the future
that is, when a retailers purchases in one period have an
effect on its purchases in subsequent periods. Given the
prevalence of this phenomenon, it is important to develop
models that account for this behavior.
We organize the remainder of this article as follows: In
the next section, we lay out the basic model and detail our
assumptions. In subsequent sections, we analyze forward
buying within the three different channel structures. Then,
we allow for trade promotions and explore the impact of
forward buying on trade promotion.
MODEL
We begin with the simplest possible model that can cap-
ture the interactions between manufacturers and retailers
and allow the retailer to forward buy. Recall that the typi-
cal reasons put forward to explain the presence of forward
buying or carrying inventory are temporary price reduc-
tions offered by the manufacturer, demand or supply uncer-
tainty, demand or supply lead times, and retailer ordering
costs. Our initial model specically rules out these rea-
sons; thus, there is no temporary price cut, no uncertainty,
no lead times, and no ordering costs. By ruling out these
reasons, we can determine whether there is an alternative
92 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2010
explanation for forward buying and isolate the effect of for-
ward buying on manufacturers and retailers. Subsequently,
we incorporate trade promotion and retailer merchandising
effort into our analysis.
We assume a two-period model in which the retailer has
the option to purchase additional units in Period 1 and
carry them as inventory into Period 2. Because there is
no uncertainty about demand, a decision to forward buy is
based solely on the wholesale prices offered by the manu-
facturer. We analyze these wholesale prices both with and
without trade promotions. We rst analyze cases in which
the retailer has the option to forward buy when the man-
ufacturer does not offer any trade promotions. Then, we
consider the case in which the retailer has the option to for-
ward buy during a trade promotion. Therefore, the retailer
in our model may forward buy even when trade promotions
are not offered and may forgo forward buying even in the
presence of trade promotions.
We begin our analysis with a base model in which one
manufacturer sells its products through a single retailer.
Subsequently, we examine two duopoly cases, one with two
competing manufacturers selling through a single retailer
and the other with a single manufacturer selling through
two competing retailers. Finally, we consider a case in
which the manufacturer offers a trade promotion to its
retailer. In the remainder of this section, we describe the
base model with a single manufacturer selling through a
single retailer. In the subsequent sections, we describe the
embellishment to the basic structure that we consider.
Consumer demand for the product in period t (t = 1. 2),
d
t
, is given by
d
t
= . p
t
. (1)
where . is the base level of demand for the product and p
t
is the retailers price in period t.
In each period, the manufacturer offers the retailer the
opportunity to purchase goods at an announced wholesale
price, w
t
. We relax a major assumption of prior models and
distinguish between the retailers order quantity and selling
quantity. Thus, given the wholesale price, the retailer must
make two simultaneous decisions in each period: the quan-
tity of units to order from the manufacturer and the retail
price to charge consumers.
2
The forward-buying quantity
or the inventory in period t (t = 1. 2), I
t
, is the difference
between the quantity q
t
that the retailer orders from the
manufacturer and the quantity d
t
that consumers demand at
the price chosen by the retailer: I
t
= q
t
d
t
0. If the retailer
carries inventory, it incurs a holding cost of h > 0 per unit.
Units carried in inventory do not deteriorate or perish and
can be sold in the subsequent period as new goods. The
manufacturer faces a constant marginal cost of production
that we set to zero. Both players face the same discount
factor, o (0. 1].
In each period, there are two stages. In the rst stage,
the manufacturer makes its wholesale price decision, and in
the second stage, the retailer makes its order quantity, retail
price, and merchandising decisions. Thus, the four stages
of the game are as follows:
2
We can generate qualitatively identical results when the retailers deci-
sions are made sequentially. However, the algebra is more tedious and the
intuition is less clear in that case.
Stage 1: The manufacturer sets the rst period wholesale
price, w
t
.
Stage 2: The retailer chooses the rst period order quantity,
q
1
, and the rst period retail price, p
1
.
Stage 3: The manufacturer sets the second period wholesale
price, w
2
.
Stage 4: The retailer chooses the second period order quantity,
q
2
, and the second period retail price, p
2
.
We adopt the notion of subgame perfect Nash equilib-
rium and solve the game backward, starting from Stage 4.
In the subsequent sections, we solve this game in several
channel settings that differ on the number of players that
are in competition. In the simplest case, we study forward
buying in a situation in which a single manufacturer sells
to a single retailer. Subsequently, we introduce competi-
tion, rst at the manufacturer level and then at the retailer
level.
ONE MANUFACTURERONE RETAILER CHANNEL
The simplest possible case to analyze is one in which a
single manufacturer sells through a single retailer. Anand,
Anupindi, and Bassok (2008) also analyze this case under
the assumption that there is no discounting. In this section,
we consider the more general case in which the retailer and
the manufacturer face a discount factor, o (0. 1].
3
This
serves as a useful benchmark for the subsequent sections
when we consider competition and the impact of trade pro-
motions and demand uncertainty.
We begin with the analysis of the retailers second period
(Stage 4) decisions. In Period 2, the retailer has I
1
0 units
in its inventory, and thus the maximum number of units it
can sell is q
2
+ I
1
. At this stage, there is no reason to have
any unsold units at the end of the period. Therefore, the
actual sales at price p
2
is going to be the smaller of two
quantities: . p
2
and q
2
+ I
1
. Thus, the retailers optimiza-
tion problem is given by
Max
q
2
. p
2
t
R
2
= p
2
(min]. p
2
. q
2
+I
1
]) w
2
q
2
.
where t
R
2
is retailer Rs prots in Period 2. At a given price
p
2
, the retailers optimal ordering quantity is q

2
= . p
2
I
1
.
Substituting q

2
in t
R
2
and solving the optimization problem,
we get the following:
p

2
=
. +w
2
2
. q

2
=
. w
2
2
I
1
.
This shows that as the retailer forward buys more units in
Period 1, it orders fewer units in Period 2.
The manufacturers prot function at this stage is
given by
t
M
2
= w
2
q

2
= w
2

. w
2
2I
1
2

.
3
The results in Anand, Anupindi, and Bassok (2008) can be easily repli-
cated by setting o = 1 in this section. Note that the full cost of forward
buying is captured not only through the holding cost but also through the
discount factor. As we show subsequently, both parameters play a crucial
role in determining the effects on the manufacturer and retailer.
Forward Buying by Retailers 93
Anticipating the retailers decision, the manufacturer max-
imizes its Period 2 prot by choosing its optimal Period 2
wholesale price:
w

2
=
.
2
I
1
.
This shows that as the retailer forward buys more units
in Period 1, it decreases its optimal ordering quantity in
Period 2, forcing the manufacturer to decrease its Period 2
wholesale price.
Next, we analyze the Period 1 decisions, beginning with
the retailers decisions in Stage 2. The retailers prot in
Period 1 is t
R
1
= p
1
(. p
1
)w
1
q
1
hI
1
. The retailer chooses
q
1
and p
1
to maximize the discounted sum of its prots
over the two-period horizon,
H
R
= t
R
1
+ot
R
2
= p
1
(. p
1
) w
1
q
1
hI
1
+ot
R
2
.
The optimal values of q
1
and p
1
are given by
p

1
=
. +w
1
2
. q

1
= Max

o(6. 3w
1
) 4(h +w
1
)
6o
.
. w
1
2

.
Given the retailers optimal choices, the manufacturer
chooses the Period 1 wholesale price to maximize the dis-
counted sum of its prot, H
M
= w
1
q

1
+ot
M

2
. This yields
w

1
=
9o. 2h
8+9o
when 0 h < h
r1
=
.o(9o4)
8+12o
and w

1
=
.
2
otherwise.
This leads to the following proposition:
4
P
1
: The retailer forward buys if and only if 0 h < h
r1
. How-
ever, the manufacturer is better-off with the retailers for-
ward buying only when
0 < h < h
m1
=
.|2o(1o)

o(8+9o)]
4(1+o)
< h
r1
.
P
1
highlights an important result: When the holding cost
is not too high, the retailer orders more units than it plans to
sell in Period 1, holds the additional units in its inventory,
and sells them in Period 2. This happens in our model in
the absence of all the typical reasons for a retailer to carry
inventorynamely, demand or supply uncertainty, supply
lead times, or high ordering costs. Forward buying occurs
because a positive inventory in Period 2 gives the retailer a
strategic advantage that leads the manufacturer to charge a
lower wholesale price in Period 2. Although forward buy-
ing in Period 1 clearly has a benet in Period 2, it also
has additional holding costs in Period 1 and the possibility
that the manufacturer can raise wholesale price in Period 1.
These additional costs in Period 1 can offset Period 2 ben-
ets of forward buying to the retailer. Therefore, forward
buying is not always optimal for the retailer but is optimal
when the holding costs are sufciently low, 0 h < h
r1
.
Conventional wisdom in marketing argues that manufac-
turers are hurt by forward buying by retailers. We acknowl-
edge that retailers forward buying can have other negative
4
All proofs are available in the Web Appendix (http://www.marketing
power.com/jmrfeb10).
effects that are not captured in our model (e.g., variable
production cycles that increase manufacturing costs), but
P
1
shows that forward buying by a retailer can have a pos-
itive impact on the manufacturers prots. Therefore, even
when the manufacturer is able to prevent forward buying
by the retailer, it may choose not to do so. However, when
0 < h
m1
< h < h
r1
, the manufacturers prot decreases with
forward buying, whereas the retailers prot increases with
forward buying. We summarize these ndings in Figure 1.
Note that when there is no discounting, h
r1
= h
m1
and both
the retailer and the manufacturer are always better-off with
forward buying.
Consider how the retailers optimal ordering quantity
(q
1
) as a function of wholesale price (w
1
) changes with
forward buying. In particular, the retailers optimal q
1
with
and without forward buying is given by
q

1
=
(6.o4h) (3o+4)w
1
6o
if I

1
> 0. and
q

1
=
. w
1
2
if I

1
= 0.
Thus, when the retailer forward buys, any change in
the Period 1 wholesale price has a greater impact on the
retailers purchase quantity. Because of this, the retailer not
only shifts part of its Period 2 purchase to Period 1 but
also increases the total quantity it purchases across the two
periods. In some cases, this increase in total quantities also
increases the manufacturers prots.
Although forward buying in this framework is driven by
wholesale prices charged by the manufacturer, the relation-
ship between the two wholesale prices is not completely
straightforward. In particular, from Table 1,
w

1
w

2
=
3.o(3o2) 4h(1+2o)
o(8+9o)
.
Therefore,
w

1
w

2
h h
w
=
3to(3o2)
4+8o
.
Because h
w
0 for any value of o 23, w

1
< w

2
for any
positive value of holding costs so long as o 23. When
o > 23, w

1
can exceed w

2
only when the holding costs are
sufciently low. If we assume that there is no discounting
(as in Anand, Anupindi, and Bassok 2008), then w

1
w

2
.
Our results show that though the presence of a positive
discount rate is not necessary for the retailer to forward
buy, the discount rate determines the wholesale price path,
which depending on the rate can be either decreasing or
increasing over time.
COMPETITION AND FORWARD BUYING
In this section, we study two cases that explore the effect
of competition on forward buying. In the rst, we consider
a single manufacturer selling to two competing retailers,
and in the second, we consider two competing manufactur-
ers selling through a single retailer.
Two RetailersOne Manufacturer
We modify our base model to allow for two retailers,
A and B, which sell a product from a single manufacturer.
94 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2010
Figure 1
CUTOFF VALUES OF HOLDING COST
Manufacturer
and retailer
better-off with
forward buying
Retailer better-off with
forward buying
0
.2 .6 .8 1.0
h
r1
h
m1
Both manufacturer and
retailer worse-off with
forward buying
h

.4
Retailers A and B are symmetric and differentiated from
each other, and their demand functions are given by
d
At
=
1
2
|. p
At
+0(p
Bt
p
At
)]
d
Bt
=
1
2
|. p
Bt
+0(p
At
p
Bt
)]

. (2)
where the parameter 0 represents the intensity of competi-
tion between the two retailers. These demand functions are
based on the quadratic utility function developed by Shu-
bik and Levitan (1980) and are analogous to the demand
function we used in the previous section (see Equation 1).
Note that the parameter 0 applies only when both demands
are positive. An appealing property of this formulation is
that the intercept for the total demand does not change as
a consequence of bringing an additional manufacturer into
the market. This ensures that if we observe forward buy-
ing in this framework, it is not because of an expansion of
demand that may arise from a second retailer entering the
market. Because both retailers are symmetrical, the man-
ufacturer cannot discriminate between them and charges
them the same wholesale price.
The sequence of events is the same as before except
that the two retailers make their price and ordering quan-
tity decisions simultaneously. Therefore, we report only
the important parts of the analysis and delegate the details
to the Web Appendix (http://www.marketingpower.com/
jmrfeb10). The Period 2 optimal price and ordering quan-
tity for Retailer A are as follows (Retailer Bs decisions are
symmetrically dened):
p

A2
=
. +w
2
(1+0)
2+0
. q

A2
=
(. w
2
)(1+0) 2(2+0)I
A1
2(2+0)
. (3)
As in the previous case, if a retailer carries inventory
from the previous period, it buys less in Period 2. An
important effect of competition is that as the competitive
intensity increases, each retailers price responds more to
Table 1
ANALYSIS OF A SINGLE MANUFACTURER AND A SINGLE
RETAILER CHANNEL
Condition 0 h < h
r1
=
.o(4+9o)
8 +12o
h h
r1
0
w
1
9.o2h
8+9o
.
2
w
2
2
3o(h +.) +h
o(8+9o)
.
2
q
1
.o(4+9o) 2h(4+5o)
2o(8+9o)
.
4
q
2
3o(h +.) +2h
o(8+9o)
.
4
d
1
4. +h
8+9o
.
4
d
2
.o(2+9o) 2h(2+3o)
2o(8+9o)
.
4
p
1
.(4+9o) h
8+9o
3.
4
p
2
.o(14+9o) +h(4+6o)
2o(8+9o)
3.
4
I
1
.o(4+9o) 4h(2+3o)
2o(8+9o)
0
Forward Buying by Retailers 95
the wholesale price charged by the manufacturer. More
formally,
cp

A2
cw
2
=
1+0
2+0
> 0. and
c
2
p

A2
cw
2
c0
=
1
(2+0)
2
> 0.
We know from the previous discussion that the main ben-
et of forward buying for the retailer is that it enjoys a
reduction in Period 2 wholesale price. However,
c
2
p

A2
cw
2
c0
> 0.
indicates that with competition, a greater part of any reduc-
tion in Period 2 wholesale price will get passed on to the
customers, and therefore the retailer may have less to gain
from such reductions in wholesale price.
5
To better understand why retailers may have less to gain
from wholesale price reductions, consider how the retailers
Period 2 prot is affected by changes in the Period 2 whole-
sale price. In particular,
ct

A2
cw
2
=
(. w
2
)(1+0) +2I
A2
(2+0)
2
(2+0)
2
< 0. and
c
2
t

A2
cw
2
c0
=
(. w
2
)0
(2+0)
3
> 0.
In other words, the retailers Period 2 prots increase with
a decline in the Period 2 wholesale price, but this change
becomes smaller as the competition between the retailers
increases.
The manufacturers optimal wholesale price in Period 2
is given by
w

2
=
.(1+0) (I
A2
+I
B2
)(2+0)
2(1+0)
. (4)
Equation 4 shows two new strategic effects that are due
to the retail competition. First, when either retailer carries
inventory from the previous period, the Period 2 wholesale
price decreases. Therefore, even if a single retailer car-
ried inventory from Period 1, the manufacturer reduces the
Period 2 wholesale price for both retailers. This results in a
free-riding problem between the two retailers: Each retailer
wants the benets of a lower w
2
but may have an incen-
tive to let the other retailer carry the inventory and incur
the holding costs. Second, compared with the monopoly
case, for a given level of inventory a retailer carries, the
manufacturers wholesale price is less sensitive to changes
in retailer inventory. This arises because the competition
between the retailers dilutes each ones market power.
Given the optimal prices and quantities in Period 2, each
retailer maximizes its overall prots by making its Period 1
price and ordering quantity decisions. Here, we provide the
optimal choices for Retailer A and note that Retailer Bs
choices are symmetric:
p

A1
=
.(10+60+o0
2
) +2(1+0)|w
1
(5+20) 0h]
20+220+60
2
. (5)
(6)
q

A1
=
2(1+0)o|5.(2+0)+0hw
1
(5+20)].0
2
o
2
8(h+w
1
)(1+0)(2+0)
4o(2+0)(5+30)
.
5
Desai (2000) observes a similar effect in a single period model.
Similar to Period 2, as the competition between the retailers
becomes more intense, the retail price is more sensitive to
changes in the wholesale price.
Table 2 shows the equilibrium choices of all the players
and demonstrates that even in the case of retail competi-
tion, the retailers may engage in forward buying. The next
proposition describes how the extent of forward buying is
inuenced by the intensity of competition.
P
2
: As the competition between the retailers increases, each
retailer decreases its equilibrium forward-buying quantity.
This result arises because of the effects described previ-
ously: Compared with the monopoly case, an increase in
inventory leads to a smaller reduction in wholesale prices.
Furthermore, a wholesale price reduction is less valuable
for the retailer because a larger fraction of it needs to be
passed on to nal consumers. Finally, each retailer has
incentives to free ride on the forward buying done by the
other retailer.
An implication of this result is that there are conditions
under which a retailer would forward buy in a less com-
petitive situation but would not do so in a more competi-
tive situation. This suggests that price competition among
retailers can be exacerbated by forward buying: If retailers
carry inventory in a highly competitive market, they also
have an incentive to lower retail prices in both periods and,
as a consequence, earn lower prots. This raises the poten-
tial for retailers to engage in forward buying because they
might nd themselves in a prisoners-dilemma situation.
This leads to the following proposition:
P
3
: When 0 < h < h
r3
, both retailers nd it optimal to forward
buy. However, when h
r4
< h < h
r3
, both retailers are worse-
off with forward buying than when neither one forward
buys.
P
3
conrms our conjecture of a prisoners dilemma, and
we nd that even though the two retailers may be worse-off
with forward buying, they still forward buy for competitive
reasons. Finally, we conclude this section by noting that
as in the previous case, for some values of the parameters,
the manufacturer can also be better-off with the retailers
forward buying.
Two ManufacturersOne Retailer Channel
Now we consider our second case of competition, specif-
ically the effect of manufacturer competition on the inci-
dence and protability of forward buying by a retailer. We
consider two manufacturers selling to a single retailer and
modify our demand function as follows:
d
it
=
1
2
|. p
it
+(p
jt
p
it
)]
d
jt
=
1
2
|. p
jt
+(p
it
p
jt
)]

. (7)
where i and j denote the two manufacturers, is a param-
eter representing the intensity of competition between the
two manufacturers (or the substitutability between the prod-
ucts), and t (t = 1. 2) denotes time. These demand functions
are analogous to the demand function used in the previ-
ous section (i.e., the intercept for the total demand for the
goods is xed). That is, compared with the previous sec-
96 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2010
Table 2
ANALYSIS OF A SINGLE MANUFACTURER AND TWO-RETAILER CHANNEL
0 h < h
r3
= o.
]25 +0|45 +0(26 +50)]] 2(1+0)(2+0)(4+30)
2(1+0)(2+0)
2
(4+5o)
h h
r3
0
w
A1
2(1+0)|h0(5+30) +.(2+0)(25+170)]o.(0o)
2
(5+30) 8h(1+0)
2
(2+0)
4(1+0)]24+25o+0|28+25o+0(8+6o)]]
.
2+0
w
A2
2h(1+0)(2+0)
2
(4+5o) +.o]2(1+0)(2+0)(10+70) +0|5+0(5+0)]o]
2o(1+0)]24+25o+0|28+25o+0(8+6o)]]
.
2+0
q
A1
.o|8(1+0)(2+0)
2
+2(1+0)(2+0)(25+80)o(0o)
2
(5+20)] +2h(1+0)|0o
2
(5+20) 12o(2+0)(3+0) 16(2+0)
2
]
8o(2+0)]24+25o+0|28+25o+0(8+6o)]]
.(1+0)
4(2+0)
q
A2
2h(1+0)(2+0)
2
(4+5o) +.o]2(1+0)(2+0)(10+70) +0|5+0(5+0)]o]
4o(2+0)]24+25o+0|28+25o+0(8+6o)]]
.(1+0)
4(2+0)
d
A1
.|96+0(2(104+o) +0](245o)(6+o) +20|16o(2+o)]])] +2h(1+0)|8+0(5+20)(4+o)]
8(2+0)]24+25o+0|28+25o+0(8+6o)]]
.(1+0)
4(2+0)
d
A2
.o]4+25o+0|6+35o+0(2+11o)]] 2h(1+0)(2+0)(4+5o)
4o]24+25o+0|28+25o+0(8+6o)]]
.(1+0)
4(2+0)
p
A1
.|96+200o+0(112+298o+0]32+o|154+5o+20(14+o)]])] 2h(1+0)(2+0)|8+0(5+20)(4+o)]
4(2+0)]24+25o+0|28+25o+0(8+6o)]]
.(3+0)
2(2+0)
p
A2
2h(1+0)(2+0)(4+5o) +.o]44+25o+0|50+15o+0(14+o)]]
4o(2+0)]24+25o+0|28+25o+0(8+6o)]]
.(3+0)
2(2+0)
I
A1
.o|25o+0
3
(5o6) +90(5o4) 16260
2
(1o)] 2h(1+0)(2+0)
2
(4+5o)
2o(2+0)]24+25o+0|28+25o+0(8+6o)]]
0
Notes: Retailer Bs decisions are dened symmetrically.
tion, in which the retailer sells a single product, by adding
another manufacturers product to its line, the retailer does
not expand the size of the market. Furthermore, if retailer
sells zero units of a product, the demand system reverts to
the single product case (Equation 1), and the idea of substi-
tutability () between the two products is moot.
6
Finally, it
is always optimal for the retailer to sell both manufacturers
products.
The manufacturers are symmetrical in all respects and
move simultaneously to choose their wholesale prices. The
other aspects of the model are the same. Because we solve
the model in a manner that is similar to the procedure used
in the previous section, we do not present all the details.
In Period 2, the retailer maximizes prots by choosing the
optimal quantity to order and retail price to charge. This
yields the following:
p

i2
=
. +w
i2
2
. q

i2
=
. (1+)w
2i
+w
2j
4I
2i
4
. (8)
The retailers decisions for Manufacturer j are symmet-
rically dened. The two manufacturers maximize their
Period 2 prots by simultaneously choosing their optimal
wholesale prices. This yields the following:
w

i2
=
.(2+3) 8I
i1
(1+) 4I
j1

4+8+3
2
. (9)
w

j2
=
.(2+3) 8I
j1
(1+) 4I
i1

4+8+3
2
.
6
If continued to play a role with zero units of one of the products, we
would have a perverse case of a money pump, in which the manufacturer
sets an exorbitantly high price for one product to drive up demand for the
other.
These equations show the following two effects of the
retailers forward buying on the manufacturers.
1. Direct effect: When the retailer has inventory of manu-
facturer is product, the retailer buys less from manufacturer
i, which leads the manufacturer to lower its optimal whole-
sale price.
2. Strategic effect: When the retailer has manufacturer js prod-
uct in its inventory, it leads manufacturer i to lower its
wholesale price.
Note that the direct effect is similar to the effect in the
bilateral monopoly case. The strategic effect arises because
of the competition between the two manufacturers and
results in each manufacturers wholesale price declining
with the competing manufacturers price.
The retailers optimal Period 1 decision for manufacturer
is products are as follows (its decisions for manufacturer
js products are symmetrically given):
p

i1
=
. +w
i1
2
. q

i1
=
.
2
+
.
j
w
j1
.
i
w
i1
h(2+)
2
(3+4)
4(3+2)(3+4)o
. (10)
where .
i
= (1 + )|12 + 9o + (24 + 11+ 18o + 8o)] and
.
j
= |8+9o+(16+7+18o+8o)].
As in the monopoly manufacturer case, a manufacturers
choice of Period 1 wholesale price affects not only the
retailers Period 1 decisions but also its Period 2 deci-
sions. Furthermore, with forward buying, the retailer orders
a higher total quantity than it would order if there were no
forward buying. In addition, because of the substitutability
between the two products, any increase in one manufac-
turers wholesale price leads the retailer to shift demand
toward the competing manufacturer. As a result, even when
the retailer engages in forward buying, the competition
Forward Buying by Retailers 97
between the two manufacturers limits each manufacturers
ability to increase its rst period wholesale price.
We provide the full solution to this game in Table 3.
Note that the manufacturers optimal wholesale prices in
Period 1 are given by
w

i1
= w

j1
(11)
=
2.(3+2)
2
(3+4)oh(2+)(6+92
3
)
48+156+186
2
+99
3
+20
4
+(2+)(3+2)
2
(3+4)o
.
This leads to the following proposition:
P
4
: The retailer nds it optimal to forward buy from two com-
peting manufacturers when
0h<
.o|(2+)(3+2)
2
(3+4)o246040
2

3
+4
4
]
(2+)
2
|(2+)(3+2)(3+4)o+12+36+35
2
+11
3
]
.
As in the manufacturermonopoly case, the retailer nds
it optimal to buy more than what it needs in Period 1 so it
can get lower wholesale prices in Period 2. The main dif-
ference here is that forward buying of either manufacturers
product lowers the wholesale price of both manufacturers
products in Period 2. Therefore, when there is competition
between the manufacturers, forward buying in Period 1 pro-
vides greater Period 2 benets to the retailer. Essentially,
forward buying of either product allows the retailer to play
the manufacturers off each other.
This discussion also indicates that the manufacturers
may have less to gain from the retailers forward buying
when there is competition between the manufacturers. We
can also derive the conditions in which the manufacturers
are better-off with the retailers forward buying. However,
intractable algebra prevents us from fully characterizing the
Table 3
ANALYSIS OF TWO MANUFACTURERS AND A SINGLE RETAILER CHANNEL
0 h < h
r2
= o.
|o(2+)(3+2)
2
(3+4) +4
4

3
40
2
6024]
(2+)
2
|12+36+35
2
+11
3
+o(2+)(3+2)(3+4)]
h h
r2
0
w
i1
2.o(3+2)
2
(3+4) h(2+)(6+92
3
)
48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)
.
2(2+0)
w
i2
(2+)]o(3+2)(3+4)|2. +h(2+)] +h(1+)|12+(24+11)]]
o|48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
.
2(2+0)
q
i1
(1+)|2.o|12+36+37
2
+12
3
+o(3+2)
2
(3+4)] h(2+)(6(4+5o) +]60+69o+|46+44o+(11+8o)]])]
4o|48+156o+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
.(1+)
4(2+)
q
i2
(1+)(2+)]o(3+2)(3+4)|2. +h(2+)] +h(1+)|12+(24+11)]]
4o|48+1560+1860
2
+990
3
+200
4
+o(2+0)(3+20)
2
(3+40)]
.(1+)
4(2+)
d
i1
.]48+|156+186+99
2
+20
3
+o(3+2
2
)(3+4)]] +h(2+)(6+92
3
)
4|48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
.(1+)
4(2+)
d
i2
.o|12+66+118
2
+83
3
+20
4
+o(2+)(3+2)
2
(3+4)] h(2+)|11+36+35
2
+11
3
o(2+)(3+2)(3+4)]
4o|48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
.(1+)
4(2+)
p
i1
.
2
+
2.o(3+2)
2
(3+4) h(2+)(6+92
3
)
2|48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
.(3+)
2(2+)
p
i2
h(2+)
2
(3+2)(3+4) +.(84+]246+|254+5(23+4)]]) +.o
2
(2+)(3+2)
2
(3+4) +h(1+)(2+)|12+(24+11)]
2o|48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
.(3+)
2(2+)
I
i1
.o|o(2+)(3+2)
2
(3+4) +4
4
246040
2

3
] h(2+)
2
|12+36+35
2
+11
3
+o(2+)(3+2)(3+4)]
2o|48+156+186
2
+99
3
+20
4
+o(2+)(3+2)
2
(3+4)]
0
Notes: Decisions related to manufacturer j are dened symmetrically.
interaction between the degree of manufacturer competition
and forward buying. In general, it is still possible for the
competing manufacturers to be better-off with the retailers
forward buying. However, as the competition between man-
ufacturers becomes more intense (i.e., increases), the
parameter space for which this is true shrinks.
MODEL EXTENSIONS
In the previous two sections, we showed how for-
ward buying by the retailer affects manufacturer and
retailer prices and protability. Importantly, forward buying
occurred in the absence of any trade promotions offered
by the manufacturer and in the absence of any uncertainty
about demand. The question this raises is the following:
How do trade promotions and demand uncertainty affect
retailer forward buying? In this section, we focus on each
of these issues.
Incorporating Merchandising Effort and Trade Promotion
The essential idea behind trade promotions is that the
manufacturer can lower price temporarily and induce the
retailer to purchase additional units, some of which can be
carried in inventory and sold in future periods (Lal and
Villas-Boas 1998). In our basic bilateral monopoly model,
because we consider prices across two periods, the cases
in which Period 1 wholesale price is lower than Period 2
wholesale price can be thought of as a trade promotion.
Strictly speaking, a lower wholesale price in Period 1 does
not qualify as a typical trade promotion because there is no
viable alternative of a regular wholesale price. In other
words, the retailer does not have the option of rejecting the
98 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2010
manufacturers offer in favor of a regular offer. Therefore,
in this section, we formally model a trade promotion by
giving the retailer a choice between a trade promotion that
consists of a special wholesale price tied specically to a
level of merchandising effort and a regular wholesale price
with no merchandising requirement. This embellishment
enables us to examine forward buying when the manufac-
turer offers a trade promotion. Importantly, it also enables
us to examine the implications of a manufacturers policy
of disallowing forward buying (e.g., through the use of scan
backs).
We modify the bilateral monopoly model developed
previously to include merchandising effort put in by the
retailer and trade promotion offered by the manufacturer.
In particular, the demand function is given by
d
t
= . p
t
+se
t
. (12)
where e
t
is the retailers merchandising effort and s is
the effectiveness of the merchandising effort in increasing
sales. Note that the level of merchandising support offered
by the retailer increases the demand for the product. For
example, demand increases if a retailer offers more service
or provides valuable end-of-aisle displays. Finally, mer-
chandising has an in-store effect only in that period (i.e.,
merchandising effort in one period does not affect demand
in the subsequent period).
With a trade promotion, the manufacturer species the
retailers effort level, but the retailer bears the cost of mer-
chandising effort, which is given by c = e
2
2. Thus, the
trade promotion offered by the manufacturer is a reduced
Period 1 wholesale price that is tied to a specic level
of merchandising effort. The retailer has the option of
not accepting the trade promotion offer, in which case it
chooses its own optimal level of merchandising effort and
pays the regular wholesale price. More formally, when the
manufacturer offers the retailer a trade promotion, it offers
the following choice of Period 1 wholesale prices:
w
1
=

w
p
1
if e
m
e
p
m
w
r
1
w
p
1
if otherwise.
(13)
where w
p
1
is the Period 1 trade promotion price, w
r
1
is
the regular (without trade promotion) wholesale price in
Period 1, and e
p
m
is the promotional merchandising effort.
It is straightforward that the retailer would choose the pro-
motional price, w
p
1
, only if it is lower than the regular price,
w
r
1
, which also comes without any requirement on merchan-
dising effort. This formulation is consistent with industry
practice in which manufacturers offer promotional prices
that are contingent on specic performance levels. That is,
if retailers want to benet from the trade promotion, they
must also invest in additional marketing activities. Such a
performance-contingent contract has also been used by Lal,
Little, and Villas-Boas (1996), who point out that a realis-
tic model should allow the retailer the option of rejecting
a trade deal and purchasing at the regular price.
7
Finally,
7
Lal, Little, and Villas-Boas (1996) develop a model in which a man-
ufacturer offers a trade promotion to two competing retailers. In their
model, trade promotions are driven entirely by the competition for a
switching segment of consumers. In our case, the trade promotion has its
own demand-enhancing effect that is independent of any switchers in the
market.
in keeping with the spirit of trade promotions as temporary
price discounts, we assume that the retailer does not offer
a trade promotion in Period 2. Therefore, in Period 2, the
manufacturers wholesale price should be higher than the
promotional price in Period 1. In our model, this condition
is always satised when
h
3.o(3o2)
4(1+2o)
.
The manufacturer may not always nd it optimal to offer
a trade promotion, in which case the effect of forward buy-
ing is similar to our previous discussion in the bilateral
monopoly case.
8
Therefore, in this section, we focus only
on cases in which it is protable for the manufacturer to
offer a trade promotion and when it is protable for the
retailer to accept the equilibrium trade promotion offer. As
in the previous section, we let the retailer have the option
of forward buying in Period 1.
In Period 2, the retailers maximization problem is as
follows:
Max
q
2
. p
2
. e
2
t
R
2
= p
2
(min]. p
2
+se
2
. q
2
+I
1
]) w
2
q
2

e
2
2
2
.
The retailers second period choices are given by
e

2
=
s(. w
2
)
2s
2
. q

2
=
. w
2
2s
2
(q
1
d
1
).
and p

2
=
. +(1s
2
)w
2
2s
2
.
As we would expect, the optimal effort level is inu-
enced by the wholesale price the manufacturer chooses:
A higher wholesale price leads to a lower effort level. In
addition, any increase in inventory carried from Period 1
decreases the quantity ordered in Period 2.
The manufacturers second period problem is to choose
w
2
to maximize t
M
2
= w
2
q
2
, which gives the following:
w

2
=
. (2s
2
)(q
1
d
1
)
2
.
In Period 1, if the manufacturer offers a trade promo-
tion, the retailer has the option of choosing either the pro-
motional wholesale price with the manufacturer-specied
merchandising effort or the higher (regular) wholesale price
with no requirement on merchandising effort. Therefore,
the trade promotion offer must ensure that it provides the
retailer at least as much prot as the regular wholesale
price:
H
Rp
(w
p
1
. e
p
1
) H
Rr
(w
r
1
). (14)
Equation 14 is the retailers voluntary participation con-
straint for the promotion.
9
8
If we include merchandising effort, e
m
, in the one manufacturer
one retailer model, all the essential results go through with only minor
modications.
9
The model with a regular wholesale price in which the retailer also
chooses an effort level is derived similarly to the procedure laid out for
the one manufacturerone retailer model. These details appear in the Web
Appendix (http://www.marketingpower.com/jmrfeb10).
Forward Buying by Retailers 99
If the retailer chooses the trade promotion offer, its prot
in the rst period is given by
t
Rp
1
= p
1
(. p
1
+se
p
1
) w
1
q
1
hI
1

(e
p
1
)
2
2
.
The retailers problem is to choose an optimal q
1
and p
1
and
to maximize the discounted sum of its two-period prots.
Similarly, the manufacturers problem is to choose w
p
1
and
e
p
1
to maximize the discounted sum of its two-period prots.
This leads to the following:
P
5
: Both with and without a trade promotion, the retailers
optimal level of forward buying is given by
Max

0. q

1
d

1
=
3o. 4(w
1
+h)
3o(2s
2
)

. (15)
Thus, forward buying can take place both with and
without the presence of trade promotions from the manu-
facturer. Furthermore, Equation 15 shows that the level of
forward buying decreases with an increase in the wholesale
price offered by the manufacturer and that it is independent
of the specic effort level chosen by either the manufacturer
or the retailer. In addition, as the effectiveness of merchan-
dising increases, the retailer carries fewer units in inventory.
Because a trade promotion always has a lower wholesale
price, it is straightforward that the retailer will do more
forward buying if it participates in the trade promotion.
Thus, the retailer forward buys for two reasons: (1) to ben-
et from a temporary wholesale price reduction in Period 1
and (2) to create an inventory that can induce the manufac-
turer to charge a lower wholesale price in Period 2.
10
Thus,
even when we allow for trade promotions, we continue to
nd that retailer forward buying can be used to get a lower
wholesale price in the future.
P
5
also suggests that a trade promotion has a tendency
to increase retailer forward buying. This means that there
are parameters for which the retailer will forward buy only
when a trade promotion is offered. This suggests that there
can be instances in which it is not optimal for the retailer
to forward buy for strategic reasons. However, if the terms
of the trade promotion are attractive enough, the retailer
will forward buy. This result is consistent with the conven-
tional wisdom that forward buying occurs because of trade
promotions being offered by the manufacturer. Because we
analyzed forward buying without trade promotion in the
previous cases, in this section we restrict our attention to
parameters for which the retailer forward buys only with a
trade promotion.
Recall that manufacturers often complain about the neg-
ative consequences of retailer forward buying associated
with trade promotions, and some have even suggested that
they would like to eliminate the practice entirely. Eliminat-
ing forward buying smoothes out the production process
and reduces costs for the manufacturerboth these effects
are beyond the scope of this paper. However, there are other
10
The comparison here is with the Period 2 wholesale price that the
retailer would get without forward buying.
consequences of eliminating forward buying, and these are
captured by the following proposition:
P
6
: If the manufacturer disallows forward buying by the
retailer, it will need to specify a lower merchandising effort
for the trade promotion. In addition, the retailer will choose
a lower level of merchandising effort in the postpromotion
Period 2.
P
6
is applicable to the situations in which the retailer
would nd it optimal to forward buy with a trade promo-
tion. In these cases, if the manufacturer does not allow
the retailer to forward buy during a trade promotion, the
trade promotion will be less protable for the retailer. To
make the trade promotion more attractive to the retailer,
the manufacturer will need to rethink the terms of the trade
promotion such that the retailers voluntary participation
constraint is satised. The manufacturer can do this by
either requiring a lower level of costly merchandising effort
from the retailer or changing the wholesale price. What
makes this result particularly intriguing is that though the
required merchandising effect will be lower, the depth of
the promotional discount can either decrease or increase.
To understand this result, note that by disallowing for-
ward buying, the retailers rst-period order quantity, q
1
,
becomes less sensitive to the rst-period wholesale price,
w
p
1
, and more sensitive to the effort parameter, s. When
s is low, the manufacturer meets the voluntary participa-
tion constraint by reducing e
p
1
and decreasing the promo-
tion depth (i.e., increasing w
p
1
). Conversely, when s is high,
the merchandising effort has a relatively high impact on
sales, and therefore a reduction in the merchandising effort
is more costly to the manufacturer. Therefore, the manu-
facturer does not try to meet the voluntary participation
constraint solely by reducing e
p
1
but rather through a com-
bination of lower e
p
1
and an increase in promotional depth
(i.e., lower w
p
1
).
Note that disallowing forward buying in Period 1 can
carry through and have an impact in Period 2. In partic-
ular, the retailer chooses a lower postpromotion merchan-
dising effort in Period 2. The reason is that the Period 2
wholesale price is higher without forward buying, and this
decreases the marginal benets of merchandising effort for
the retailer.
Incorporating Demand Uncertainty
Uncertainty about future demand is often cited as a main
reason a retailer would hold inventory. In particular, if
demand turns out to be high, a retailer does not penalize
itself by not having enough stock on hand. In the previ-
ous sections, we showed that even when a rm is certain
about the level of demand, for strategic reasons, it may still
choose to forward buy. The issue then is to observe what
happens to forward buying when the players are uncer-
tain about demand. In particular, regardless of the demand
state, are there conditions under which a rm would hold
inventory?
We follow Desai, Koenigsberg, and Purohit (2007) and
model uncertainty by assuming that the base level of
demand, ., can be either high, . = .
H
, with probability
~, or low, . = .
L
, with probability (1 ~). The demand
state is assumed to be the same in Periods 1 and 2. When
100 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2010
rms learn the demand in Period 1, there is no remaining
uncertainty about the level of demand in Period 2. There
are ve stages of the game:
Period 1
Stage 1: The manufacturer sets the Period 1 wholesale price,
w
1
.
Stage 2: The retailer chooses the Period 1 order quantity, q
1
.
After it places its order, the players learn the true
state of demand, either .
H
or .
L
.
Stage 3: The retailer sets the Period 1 retail price, p
1
, and
carries unsold units in inventory.
Period 2
Stage 4: The manufacturer sets the Period 2 wholesale price,
w
2
.
Stage 5: The retailer chooses the Period 2 order quantity, q
2
,
and the Period 2 retail price, p
2
.
As Period 1 begins, the rms observe the true level of
demand state after the retailer has ordered and the man-
ufacturer has produced the quantity for Period 1. At this
stage, there are three possible cases:
1. q
1
< d
1L
< d
1H
: The retailer does not have enough units for
either the high or the low demand states, and it does not
carry any units from Period 1 into Period 2.
2. d
1L
q
1
< d
1H
: The retailer does not have enough units for
the high demand state but has enough units for the low state.
Only when demand is low can the retailer carry inventory
into Period 2.
3. d
1L
< d
1H
q
1
: In this case, the retailer either will have just
enough units to satisfy the high demand or will carry inven-
tory into Period 2 (under both demand states).
This leads to the following proposition:
P
7
: When 0 < h < h
u2
, the retailer orders a high enough quan-
tity such that it is optimal to carry inventory regardless of
whether demand turns out to be high or low, where
h
u2
=o
.
H
|4~(4+9o)(43o)(8+9o)]4.
L
(1~)(4+9o)
4|8+9o(2+o)]
.
When h
u2
< h < h
u1
, the retailer orders a quantity such that
it is optimal to carry inventory only if demand turns out to
be low, where
h
u1
=
2~.
H
(4+3o) .
L
|4~(4o) +o(49o)]
4(2+2~ +3o)
.
When h
u1
< h, the retailer orders a quantity such that it
never carries inventory. In contrast to the foregoing results,
an integrated rm would carry inventory only in the low
demand state and never in the high demand state.
The issue here is whether the retailer carries inventory
or not. Carrying inventory in the high demand state is evi-
dence of strategic behavior. Most models ignore this strate-
gic effect because they assume that the retailer would never
carry inventory in the high demand state. However, we nd
that under certain conditions, the retailer will carry inven-
tory not only in the low but also in the high demand state.
Finally, a way to understand our results is to compare
the decisions made by an integrated rm facing uncer-
tain demand with the retailers decisions. In this case,
facing uncertain demand, the integrated rm views inven-
tory solely as a safety stock. In contrast, because of the
manufacturers problem with time consistency, a retailer
views inventory as a safety stock and a strategic variable.
From this perspective, we nd that unlike the retailer, an
integrated rm would never carry inventory in the high
demand state.
DISCUSSION AND CONCLUSION
We began this article by noting that the conventional
wisdom in marketing is that trade promotions are the lead-
ing culprits behind retailer forward buying (e.g., Kotler
and Keller 2006). Moreover, if there is no overall increase
in consumer demand associated with the trade promo-
tion, then all the manufacturers have achieved is to sell a
larger quantity at a lower pricea practice that helps the
retailer at the expense of the manufacturer. Furthermore,
manufacturers must deal with the costs of large swings in
production volume, which leads to a further decrease in
prots (Ailawadi, Farris, and Shames 1999). In contrast,
Anand, Anupindi, and Bassok (2008) show that manufac-
turers are always better-off by allowing retailers to for-
ward buy when the manufacturers cannot commit to future
prices. In this article, we examine additional complexities
that arise from forward buying in a variety of settings,
including product-market competition, demand uncertainty,
and trade promotions.
We nd that regardless of whether trade promotions are
offered by a manufacturer, forward buying can still be
an optimal strategy for a retailer. Thus, a trade promo-
tion increases the level of forward buying, but eliminating
trade promotions does not mean that forward buying will
go away. Importantly, however, forward buying need not
always be protable for retailers. In particular, when two
competing retailers purchase from a single manufacturer,
competition forces each retailer to pass through more of the
wholesale price reductions. In some cases, retailers forward
buy because they may be in a prisoners-dilemma situation.
This result provides a contrast to the belief that though
manufacturers are hurt by retailers forward buying, they
allow this practice because of competitive pressures from
other manufacturers. Our analysis suggests that the level of
retailer competition can also drive forward buying.
The main effect of forward buying is that it results in
an overall increase in the purchase order from the retailer.
In contrast to the commonly held belief that the manufac-
turer is hurt by forward buying, we show that the manufac-
turer is better-off when the increase in its total sales offsets
the reduction in wholesale prices. Furthermore, when two
competing manufacturers are selling to a common retailer,
forward buying becomes even more likely than the man-
ufacturer monopoly case. However, competition between
manufacturers limits each ones ability to capitalize on the
forward buying.
Although conventional wisdom argues that eliminating
forward buying would be good for manufacturers, we show
that this logic holds only when the holding costs are high
enough. For lower levels of holding cost, forward buy-
ing by the retailer can benet the manufacturer and the
overall channel. In other words, forward buying potentially
can move the channel closer to a coordinated level. The
Forward Buying by Retailers 101
essential problem in channel coordination is that retailers
and manufacturers have conicting incentives that increase
channel inefciencies and lower the overall prots of the
channel.
11
Our analysis suggests that by lessening the
problem associated with double marginalization, forward
buying can play a previously undiscovered role in moving
the channel closer to a coordinated level.
12
The essential
problem with double marginalization is that, compared with
an integrated channel, the total quantity sold is too low.
Conversely, forward buying leads to a decrease in the aver-
age wholesale price charged by the manufacturer and an
increase in the total quantity sold, thus bringing the solution
closer to that of an integrated channel. Furthermore, our
analysis suggests that, under certain conditions, total chan-
nel prots can increase with holding costs. This surprising
result occurs only when the holding cost inefciency is off-
set by a decrease in the double marginalization inefciency.
Disallowing forward buying during a trade promotion
can have other consequences as well. In particular, we
nd that if a retailer is disallowed from forward buying,
the manufacturer will need to reduce the merchandising
requirement associated with the trade promotion. Further-
more, in the subsequent period when there is no promo-
tion, the retailer will also reduce its own merchandising
effort. These decreases in overall effort result in an overall
decrease in demand for the product. The effect of disal-
lowing forward buying on the trade promotion wholesale
price depends on the effectiveness of the merchandising
effort. In particular, when merchandising has a large effect
on demand, the manufacturer must increase the depth of
the trade promotion; conversely, when merchandising has
a small effect on demand, the manufacturer must decrease
the depth of the promotion. All these changes can have
important effects on the manufacturers protability.
Consider the role of forward buying in the context of
demand uncertainty. The typical way to model uncertainty
is that if a retailer is uncertain about whether demand will
be high or low, it keeps inventory on hand. Note that
inventory plays the role of safety stock; that is, it provides
protection for the case when demand turns out to be high
and does not particularly help if demand turns out to be low.
Thus, it is typically assumed that the rm will not order
more than what it expects to sell in the high demand state.
In this article, we relax that assumption and nd that in
some cases, the retailer ends up holding inventory regard-
less of whether demand turns out to be high or low. This
11
A large portion of channels research is focused on understanding
the nature of these inefciencies and on designing contractual and non-
contractual mechanisms that align manufacturers and retailers incen-
tives, such that retailers choose the appropriate price, service, promotional
spending, or any other marketing decision (see Bruce, Desai, and Staelin
2005; Coughlan and Wernerfelt 1985; Cui, Raju, and Zhang 2007; Desai,
Koenigsberg, and Purohit 2004; Lal 1990; Lee and Staelin 1997; McGuire
and Staelin 1983; Moorthy 1987). The economics literature on double
marginalization (Spengler 1950), downstream moral hazard (Holmstrom
1979), and bilateral moral hazard (Bhattacharya and Lafontaine 1995;
Desai 1997; Holmstrom 1982) also deals with vertical relationships in
which incentives are misaligned.
12
It is well established that a linear price contract, such as a wholesale
price charged by the manufacturer, leads to inefciencies because of dou-
ble marginalization. Conversely, in many situations, including in some of
the cases we consider in this article, more general contracts can lead to
full channel coordination.
is noteworthy and further highlights the role of strategic
forward buying by the retailer.
Our research adds to the extensive literature on distribu-
tion channels by discovering a new insight from a famil-
iar framework. In particular, Spenglers (1950) bilateral
monopoly model has been used extensively in marketing,
beginning with McGuire and Staelin (1983). With a few
exceptions, most applications of this manufacturerretailer
framework have been in static settings and have assumed
that the retailers ordering quantity is also its selling quan-
tity. The fundamental problem we pose occurs over time:
Forward buying by the retailer in one period has an effect
in the subsequent period, and we need to separate the order-
ing and selling decision. When we take a simple channel
model and introduce a linkage over time, it gives us a new
perspective on the basic model. Allowing forward buying
in a one-level marketing channel signicantly changes the
nature of the solution because it alters the strategic inter-
action between the manufacturer and the retailer. In par-
ticular, in the standard static framework, the manufacturer
plays the role of Stackelberg leader and is able to take
advantage of its position by moving rst. In our dynamic
setting, the manufacturer still plays the role of leader in
Period 1, but the presence of retailer inventory in Period 2
takes away part of the manufacturers advantage. Another
way of thinking about this is that by having inventory in
Period 2, the retailer needs the manufacturer only for the
residual demand that its inventoried units cannot satisfy; in
this way, the retailer can be thought of as playing the role
of leader in Period 2 and earning higher prots than before.
As we show, depending on the parameters, the manufac-
turers protability is both helped and hurt by this behavior.
To maintain tractability, we needed to make some sim-
plifying assumptions. In particular, we made the choice
of linear demand function for this reason. However, Lee
and Staelin (1997) show that in many channel models,
the nature of strategic interaction (strategic substitutability
versus strategic complementarity) rather than the specic
demand function determines the equilibrium outcomes.
Despite using a relatively simple demand function, we
were not able to analyze a model in which both retailers
and manufacturers faced competition. However, our analy-
sis provides insights into the individual effect of competi-
tion at each level. In other words, although we are unable
to determine much about the interaction between the two
types of competition, we can describe their main effects.
Finally, we acknowledge that having only two periods may
seem to be a limiting assumption. Dynamic models must
often make this assumption for tractability reasons (see
Hauser, Simester, and Wernerefelt 1994). We also analyzed
a more general n-period version of the one manufacturer
one retailer model and found that depending on the level of
holding costs, the retailers carry inventory in some periods.
There are several avenues for extending our research.
One possibility is to allow the manufacturers to charge
quantity discounts or quantity premiums. This can allow
the manufacturers to reward or penalize forward buying as
necessary and achieve full coordination by using more gen-
eral contracts. Another factor that could affect our results,
and therefore merits further investigation, is the high-end
or low-end positioning of the retailers and manufactur-
ers. Another is to consider the situation in which both
102 JOURNAL OF MARKETING RESEARCH, FEBRUARY 2010
retailers and consumers can forward buy (Anton and Das
Varma 2005; Guo and Villas-Boas 2007; Hong, McAfee,
and Nayyar 2002). These are fruitful directions in which to
take this research.
REFERENCES
Ailawadi, Kusum, Paul Farris, and Ervin Shames (1999), Trade
Promotion: Essential to Selling Through Resellers, Sloan
Management Review, 41 (1), 8392.
, Praveen Kopalle, and Scott Neslin (2004), Predicting
Competitive Response to a Major Policy Change: Combining
Game Theoretic and Empirical Analysis, Marketing Science,
24 (1), 1224.
Anand, K., Ravi Anupindi, and Yehuda Bassok (2008), Strate-
gic Inventories in Vertical Contracts, Management Science, 54
(10), 17921804.
Anton, James and Gopal Das Varma (2005), Market Structure,
Storability and Demand Shift Incentives, RAND Journal of
Economics, 36 (3), 52043.
Bhattacharya, Sugato and Francine Lafontaine (1995), Double-
Sided Moral Hazard and the Nature of Share Contract, RAND
Journal of Economics, 26 (4), 76181.
Blattberg, Robert and Scott Neslin (1990), Sales Promotions:
Concepts, Methods and Strategies. Englewood Cliffs, NJ:
Prentice Hall.
Bruce, Norris, Preyas Desai, and Richard Staelin (2005), The
Better They Are, the More They Give: Trade Promotions
of Consumer Durables, Journal of Marketing Research, 42
(February), 5466.
Coughlan, Ann, Erin Anderson, Louis W. Stern, and Adel I.
El-Ansary (2006), Marketing Channels. Englewood Cliffs, NJ:
Prentice Hall.
and Birger Wernerfelt (1985), Credible Delegation by
Oligopolists: A Discussion of Distribution Channel Manage-
ment, Management Science, 35 (2), 22639.
Cui, Tony H., Jagmohan S. Raju, and Z. John Zhang (2007), Fair-
ness and Channel Coordination, Management Science, 53 (8),
13031314.
, , and (2008), A Price Discrimina-
tion Theory of Trade Promotions, Marketing Science, 27 (5),
77995.
Desai, Preyas (1997), Advertising Fee in Business-Format Fran-
chising, Management Science, 43 (10), 14011419.
(2000), Multiple Messages to Retain Retailers: Sig-
naling New Product Demand, Marketing Science, 19 (4),
38189.
, Oded Koenigsberg, and Devavrat Purohit (2004), Strate-
gic Decentralization and Channel Coordination, Quantitative
Marketing and Economics, 2 (1), 522.
, , and (2007), A Research Note on
the Role of Production Lead Time and Demand Uncertainty
in Marketing Durable Goods, Management Science, 53 (1),
15058.
Guo, Liang and Miguel J. Villas-Boas (2007), Consumer
Stockpiling and Price Competition in Differentiated Markets,
Journal of Economics and Management Strategy, 16 (4),
82758.
Hauser, John, Duncan Simester, and Birger Wernerfelt (1994),
Customer Satisfaction Incentives, Marketing Science, 13 (4),
32750.
Holmstrom, Bengt (1979), Moral Hazard and Observability,
The Bell Journal of Economics, 10 (1), 7491.
(1982), Moral Hazard in Teams, The Bell Journal of
Economics, 13 (2), 32440.
Hong, Pilky, Preston R. McAfee, and Ashish Nayyar (2002),
Equilibrium Price Dispersion with Consumer Inventories,
Journal of Economic Theory, 105 (2), 503517.
Kotler, Philip and Kevin H. Keller (2006), Marketing
Management, 12th ed. Englewood Cliffs, NJ: Prentice Hall.
Lal, Rajiv (1990), Improving Channel Coordination Through
Franchising, Marketing Science, 9 (4), 299318.
, John Little, and Miguel Villas-Boas (1996), Forward
Buying, Merchandising, and Trade Deals, Marketing Science,
15 (1), 2137.
and Miguel Villas-Boas (1998), Price Promotions
and Trade Deals with Multi-Product Retailers, Management
Science, 44 (7), 93549.
Lee, Eunkyu and Richard Staelin (1997), Vertical Strategic Inter-
action: Implications for Channel Pricing Strategy, Marketing
Science, 16 (3), 185207.
McGuire, Timothy and Richard Staelin (1983), An Industry
Equilibrium Analysis of Downstream Vertical Integration,
Marketing Science, 2 (2), 16191.
Moorthy, K. Sridhar (1987), Managing Channel Prots: Com-
ment, Marketing Science, 6 (4), 37579.
Narasimhan, C. (1988), Competitive Promotional Strategies,
Journal of Business, 61 (4), 42749.
Raju, Jagmohan, Seenu V. Srinivasan, and Rajiv Lal (1990), The
Effects of Brand Loyalty on Competitive Price Promotional
Strategies, Management Science, 36 (March), 276304.
Rao, Ram C. (1990), Pricing and Promotions in Asymmetric
Duopolies, Marketing Science, 36 (10), 13144.
Shubik, Martin and Richard Levitan (1980), Market Structure and
Behavior. Cambridge, MA: Harvard University Press.
Spengler, Joseph (1950), Vertical Integration and Anti-Trust
Policy, Journal of Political Economy, 58 (4), 34752.
Copyright of Journal of Marketing Research (JMR) is the property of American Marketing Association and its
content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.