This action might not be possible to undo. Are you sure you want to continue?
Vol. 55, No. 8, August 2009, pp. 1338–1346
issn00251909 eissn15265501 09 5508 1338
informs
®
doi 10.1287/mnsc.1090.1024
©2009 INFORMS
Product Variety and Endogenous Pricing with
Evaluation Costs
J. Miguel VillasBoas
Haas School of Business, University of California, Berkeley, Berkeley, California 94720,
villas@haas.berkeley.edu
O
ne important decision ﬁrms must make is to select the product line (characteristics and number of products)
to offer consumers. This paper explores the effect of the interaction between consumer evaluation costs
and pricing on the optimal product line length to offer consumers. Before deciding to buy a product among
all products offered, a consumer learns the product line length. Given the product line length, a consumer
decides whether to evaluate the products available and their prices. This decision to evaluate depends on the
expected consumer surplus after the evaluation being greater than the evaluation costs. When the ﬁrm offers
few products, the ﬁrm may not attract many consumers because of lack of product ﬁt and may be forced to offer
low prices. When the ﬁrm offers many products, all consumers will ﬁnd a great product ﬁt; that is, the variance
of consumer valuations per product chosen is lower. This allows the ﬁrm to charge high prices to extract ex
post consumer surplus, resulting in lower ex ante expected consumer surplus, which may lead consumers not
to evaluate the products in the ﬁrst place. That is, by offering fewer products a ﬁrm can commit not to extract
all possible consumer surplus. These two forces may then lead to the existence of an interior optimal number
of products to offer. The optimal number of products offered is decreasing in the evaluation costs.
Key words: marketing; product policy; pricing; search costs; information overload; consumer choice
History: Received October 16, 2007; accepted March 15, 2009, by Jagmohan S. Raju, marketing. Published
online in Articles in Advance June 1, 2009.
1. Introduction
When consumers have to incur evaluation costs prior
to making a purchase, they may refrain from product
evaluation, and purchase if they are concerned about
the ex post consumer surplus that the seller may
allow. This paper explores the possibility of a ﬁrm
committing to a reduced number of supplied prod
ucts as a way to guarantee to consumers that they will
have ex post surplus to compensate for their evalu
ation costs. There is also some evidence that when a
ﬁrm offers too many products, consumers may feel
that there are too many choices and end up not pur
chasing any product.
1
One important issue in such a
setting is that ﬁrms may also act strategically in set
ting prices and other product characteristics, which
consumers only ﬁnd out about after they incur some
evaluation costs.
This paper explores this issue of the interaction
of the ﬁrm decisions on prices and product char
acteristics to be evaluated by consumers and the
consumer evaluation costs. That is, in a choice over
load setting where consumers have evaluation costs,
what is the effect of ﬁrms being strategic about the
prices they charge? Does strategic pricing accentuate
1
Some evidence on this effect with exogenous prices can be seen,
for example, in Iyengar and Lepper (2000).
or attenuate the choice overload effects, leading the
ﬁrms to either choose a shorter or longer product line?
What is the effect of the information about prices and
qualities being observed and processed prior to con
sumers engaging in the evaluation costs of the differ
ent alternatives?
The main result of this paper is that consumers’
evaluation costs in understanding pricing and prod
uct characteristics information is a force toward a ﬁrm
offering fewer products. The idea is that by offering
fewer products, a ﬁrm can commit not to extract all
possible consumer surplus. The effect is not through
the information a product line may provide to con
sumers, but rather through the variance in consumer
valuations that it induces. If the ﬁrm offers too many
products, the existence of a product with good ﬁt for
each consumer implies lower variation in the con
sumer valuations of consumers who end up buy
ing each product, which therefore allows the ﬁrms
to extract more consumer surplus. In other words, if
there are fewer products, after being evaluated, the
products have to offer some surplus to the consumers
for them to be willing to buy, which means that the
consumers who have a good ﬁt will have a signiﬁcant
surplus. Therefore, in expectation, more consumers
will be willing to evaluate the products, which will
generate greater demand. If a seller offers too many
1338
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
Management Science 55(8), pp. 1338–1346, ©2009 INFORMS 1339
products, consumers realize that all consumers will
ﬁnd a product that provides a good ﬁt (low vari
ance of consumers valuations around best product
ﬁt), which means that consumers have to ﬁnd that
product, and the ﬁrm will extract the most it can from
the surplus generated by that good ﬁt.
This idea is present in markets where many alter
natives are offered and consumers realize that ﬁnding
the best alternative is important, as only the better
alternatives can generate some signiﬁcant surplus.
Sellers with greater variety may be able to charge
more speciﬁc (higher) prices, keeping other factors
ﬁxed, which may yield lower consumer surplus.
2
For
example, restaurants that offer greater variety in ser
vice may be able to charge higher prices for their spe
ciﬁc dishes. Supermarkets that offer a large variety
in a certain category may be able to charge higher
prices because they provide better productconsumer
ﬁt. In some markets for services, greater service is the
ability to better respond to the speciﬁc demands of
each customer (greater variety of services provided),
which can allow the ﬁrm to extract more consumer
surplus, given the speciﬁc needs satisﬁed. Examples
of some types of services where these effects may
be present are highend restaurants, hotels, or hair
dressers. Greater service in these cases may mean the
ability to respond to the speciﬁc consumer demands.
In some markets where the information on prices is
more costly to obtain or is less advertised, we may
expect that these effects are more important. The main
results here may also suggest that ﬁrms that adver
tise more (and advertise their prices) may have more
dense product lines, as the effects discussed here will
not be present.
Consider a situation when consumers are able to
observe ﬁrst the length of the product line (number of
products) offered by a ﬁrm. For example, this could
be the case when a consumer sees the length of shelf
space provided in a supermarket or department store
for a category, or when a consumer sees the total list
of products available on a webpage. In many mar
ket situations, consumers may have ﬁrst a sense of
the number of product available. Then, after observ
ing the number of products available, a consumer has
to decide whether it is worth incurring evaluation
costs to ﬁnd out the value of each product for the
consumer. Processing the information of the attributes
of each product and the prices being charged may
be costly or take time, even if the product attributes
or prices are displayed or advertised.
3
But once the
2
Note, however, that this effect holds when keeping other factors
ﬁxed. For example, if there are economies of scope in variety, sellers
that offer more variety could have lower costs, which could be a
force toward lower prices.
3
See Dickson and Sawyer (1990) on the extent to which consumers
are uninformed about prices after purchase decisions and on a
evaluation costs are incurred, they are sunk costs, and
the consumer only purchases a product if the prod
uct offers a positive surplus (excluding the evalua
tion costs, as they are sunk at the purchase stage).
The ﬁrm can then beneﬁt from reducing the surplus
of the consumers (excluding the evaluation costs) by
increasing prices or reducing product quality, if these
can be checked only after the products are evaluated.
This then gives less incentive for consumers to eval
uate the product (as they know that they will have a
reduced surplus after evaluation), and the ﬁrm ends
up with less demand. By offering a smaller number of
products, the ﬁrm can commit that it will offer some
signiﬁcant consumer surplus, as it may still ﬁnd it
proﬁtable to attract consumers who do not have such
a great ﬁt with any product (which is more likely with
a smaller number of products). With a smaller num
ber of products, the ﬁrm lowers its prices to attract the
consumers who do not have a good product ﬁt, gen
erating higher ex post surplus for the consumers who
search. That is, the interaction between ﬁrm pricing
and quality decisions and evaluation costs is a force
toward a smaller number of products.
This paper considers the pricing case and shows
that there is an interior and ﬁnite optimal number of
products. The optimal number of products is decreas
ing in the evaluation costs and converges to inﬁnity
when the evaluation costs go to zero. If the evalua
tion costs are high enough, the market can disappear.
In equilibrium, the consumers who evaluate the prod
ucts are the ones who have a higher general valuation
for the category, and some consumers who evaluate
the products end up not buying any product. The
same results (available upon request) can be obtained
when ﬁrms also make decisions on the quality of each
product, which is only found out after the consumer
evaluation, and prices are observed prior to the con
sumers incurring the evaluation costs. One can obtain
similar results when the general consumer valuation
for a category is positively correlated with how much
a consumer cares about product ﬁt.
Note that the effects of the evaluation costs pre
sented here can be seen as choice overload effects
(e.g., Iyengar and Kamenica 2007), as the existence of
evaluation costs makes the consumers choose not to
buy (and not incur the evaluation costs) when faced
with too many alternatives. In the formulation below,
evaluation costs are not allowed to vary with the
number of products offered, in order to isolate the
essence of the effect of endogenous pricing. In this set
ting, endogenous pricing leads to a small number of
conceptual model of price encoding and knowledge, and informa
tion processing with respect to prices (see also Jacoby and Olson
1977). If consumers at the time of purchase are unsure of their pref
erences at the time of consumption (Guo 2006), they may beneﬁt
less from searching and processing information.
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
1340 Management Science 55(8), pp. 1338–1346, ©2009 INFORMS
optimal products, and exogenous pricing leads to an
inﬁnite number of products. We know from previous
work (e.g., Kuksov and VillasBoas 2006) that under
exogenous pricing, if the evaluation costs are increas
ing in the number of product offered (or searched),
the optimal number of products offered may be small.
Putting the two results together, we would have that
in a setting where the evaluation costs are increasing
in the number of products offered, endogenous pric
ing would lead to an even smaller optimal number
of products offered. This issue is further discussed in
the model setup.
1.1. Literature Review
Related to this paper, there is a literature that looks
at the opportunism of sellers when consumers have
to incur sunk evaluation costs before they make a
purchase. Diamond (1971) considers search costs in
consumers ﬁnding out about prices with competing
sellers, each selling only one product, and each con
sumer having a downwardsloping demand. In con
trast, here there is only one ﬁrm selling multiple
products, and the ﬁrm can choose product attributes
and the number of products it sells. Lal and Matutes
(1994) and Wernerfelt (1994) discuss the possibility of
advertised prices as a form of the seller commiting
not to exploit the consumer sunk costs,
4
and Wern
erfelt also discusses the possibility of return policies
and seller colocation. In relation to that literature, this
paper considers the possibility of ﬁrms choosing a
product line, where the length of product line may be
more easily observed than prices or product quality.
The length of the product line commits the seller to
the extent of consumer surplus that is provided. Note
also that there may be consumer processing costs
for advertised prices, and if prices can be observed
without processing or evaluation costs, unobserved
quality decisions generate the same problems (see dis
cussion above). In addition to that literature, this
paper also notes that the product line speciﬁcation
allows ex ante evaluation costs to generate informa
tion/choice overload effects.
As noted above, some experimental work has found
that consumers are more likely to make a choice and
purchase when confronted with a smaller choice set.
This work has argued that consumers may feel over
whelmed or overloaded with too many alternatives.
This idea goes back to the information overload lit
erature (see, for example, Jacoby 1977 and the refer
ences listed there) and the idea that decision makers
may only be able to process a limited amount of infor
mation (e.g., Simon 1955, Miller 1956) and mental
4
See also Simester (1995) and Rao and Syam (2001). See also
Kuksov (2004) for the effect of search costs on product design under
competition.
evaluation costs (e.g., Bettman 1979).
5
In this regard,
Shugan (1980) considers the costs of thinking and pro
vides a quantitative measure of that cost related to the
number of comparisons necessary to make a decision,
given some level of conﬁdence. Hauser and Wernerfelt
(1990) have argued that consumers may strategically
limit their consideration sets (with search under ﬁxed
sampling) to reduce evaluation costs. Another impor
tant aspect that is not considered here involves the
possibility of using attribute information to eliminate
alternatives (Payne et al. 1988). Beach and Mitchell
(1978) address the issue of decision makers eliminat
ing options on the basis of deviation from their ideal
point. Huffman and Kahn (1998) discuss the issue of
variety choice creating confusion in preferences (see
also, Mick et al. 2004, Lurie 2004, Gourville and Soman
2005). There is also some evidence (e.g., Iyengar and
Lepper 2000) that even though lager choice sets may
reduce choice, consumers are attracted to sellers with
greater choice sets. This paper will not address this
later effect, which could depend on what is kept con
stant when choice sets are enlarged. For example, in
the particular contexts that are considered, consumers
might not have full information about search costs
and might consider larger choice sets better given the
information that they have.
Kuksov and VillasBoas (2006) consider search costs
in evaluating the ﬁt of each alternative and show
that a strategic supplier of alternatives (spanning the
alternatives through the product space) does better
by offering a ﬁnite number of alternatives than an
inﬁnite number, reducing overall consumer search
costs and leading more consumers to choose an alter
native. Some of the issues discussed in that paper
can also be seen from the side of the ﬁrms through
their communication strategies (VillasBoas 2004). Van
Zandt (2004) considers competition where ﬁrms com
municate about their products and consumers eval
uate a limited and ﬁxed number of alternatives; he
ﬁnds that there is too much communication in equi
librium, as a ﬁrm communicating about its prod
uct does not consider the negative externality on
consumer information processing affecting the other
ﬁrms. Kamenica (2008) considers the possibility that
the information implicit in the product line may affect
customers’ beliefs about how much they would enjoy
particular products.
6
Norwood (2006) considers free
entry price competition among ﬁxed products that are
5
Keller and Staelin (1987) argue that with regard to information,
quality of information may help decision making, whereas quan
tity of information may affect decision making negatively. In a
market setting, Boatwright and Nunes (2001) show that reducing
the assortment size in a grocery store can increase sales.
6
In a working paper version, Kamenica (2007) also considers en
dogenous prices with large retail cost uncertainty (unknown to the
consumers), such that prices cannot signal product characteristics.
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
Management Science 55(8), pp. 1338–1346, ©2009 INFORMS 1341
vertically differentiated, one product per ﬁrm, under
the assumption that only the most popular products
are offered; Norwood includes an approximation to
the consumer sequential evaluation process and ﬁnds
that there is excessive entry in equilibrium.
Finally, it may also be that a greater number of
alternatives leads the decision maker to delay choice
(and not choose when the choice set is ﬁrst pre
sented) in order to gather more information on the
choice problem (e.g., Dhar and Simonson 2003). This
new information could come at a lower cost than the
cost of evaluating different alternatives at the present.
For example, consumers could gain some informa
tion about what other consumers have chosen (Zhang
2009). Note that this explanation could then be seen
as a search cost explanation for no choice when many
alternatives are presented.
7
Note that the objective of
this paper is not to argue that this explanation is the
most compelling one for the existence of limited prod
uct lines.
8
In relation to the papers above, we consider the
effect of the interaction between the consumer evalu
ation costs and ﬁrm decision making about variables
that affect the consumer surplus (price and quality) on
the optimal number of products for a ﬁrm to offer. In
relation to other explanations for choice overload, this
paper focusses on the explanation based on simple
evaluation costs, which is enough to generate infor
mation overload effects with endogenous pricing.
The remainder of this paper is organized as follows.
The model is presented in §2. Section 3 presents the
main results, and §4 concludes.
2. Base Model
Consider a ﬁrm deciding how many and which prod
ucts at what prices to offer to a set of consumers.
The products are located on a circle of unit length
and have zero production costs (without loss of gen
erality). Before deciding whether to search, consumers
know their general valuation . for the category and
observe the number of alternatives n being offered
by the ﬁrm. After searching and incurring the search
costs , the consumers learn the location of all the
products offered and their prices and the location of
Wathieu and Bertini (2007) consider the effect of the price being close
to the expected willingness to pay, generating further incentives for
consumers to invest in evaluating the purchase decision.
7
Note also that consumers could prefer smaller choice sets because
of selfcontrol problems (see, for example, the discussion in Gul and
Pesendorfer 2001, Bénabou and Tirole 2004, Fudenberg and Levine
2006) and regret preferences (see, for example, Loomes and Sugden
1987, Irons and Hepburn 2007, Sarver 2008).
8
For example, the existence of ﬁxed costs per product carried can
generate a limited product line. Here, the paper presents an addi
tional force toward a smaller number of products offered.
their ideal product x.
9
The ﬁrm decides the num
ber of alternatives to offer n, the location of each of
these alternatives on the circle, and the price of each
alternative.
10
Consumers are heterogeneous on their general val
uation . for a product in the category and on the loca
tion of their ideal product x in the circle. If the product
consumed is at a distance d of the consumer’s ideal
product, the consumer incurs a disutility d, where  is
a parameter measuring the importance of the product
ﬁt. The distribution of . and x is assumed indepen
dent among consumers. The marginal distribution of
x is assumed uniform along the unit circle, and the
marginal distribution of . has positive density on the
support í0, .{, with the cumulative distribution func
tion í (.) (density 1 (.)).
The ex post utility of a consumer with valuation .
purchasing a product at price p at distance d of his
ideal product is
U =. −d −p −.
Prior to incurring the search costs , the consumer
does not know d or p but may be able to infer both the
distribution of d from the best product available and
the price p from the number of products offered n.
After incurring the search costs, a consumer chooses
to buy the best product available, which is at dis
tance d, if and only if .−d−p ≥0, as the search costs
are sunk at the purchase decision time.
A consumer with valuation . decides to incur the
search costs  if the expected surplus after search is
greater than . That is, a consumer searches if and
only if 
d
ímax{. −d −p, 0¦ n{ ≥ , where 
d
is the
expected value operator over the distance d to the best
product for that consumer. As the expected value is
increasing in ., there is going to be a threshold . such
that a consumer searches if and only if . ≥ .. The ﬁrm
chooses the prices of the different products (and their
location) to maximize proﬁts given that consumers
who have searched have . ≥ .. Note that the con
sumers foresee this behavior by the ﬁrm; therefore,
the determination of ., the valuation of the con
sumers indifferent between searching and not search
ing, takes into account the ﬁrm’s pricing strategy.
9
Consumers could also have search costs per additional product
being evaluated, but the main messages of the results presented
here should also extend to that setting. That setup would compli
cate the analysis, although it would not be central to the market
forces presented here. Search costs could also be different across
consumers without affecting the main results presented here.
10
Note that if the search costs  are increasing in the number of
products offered n, we would then immediately obtain choice over
load effects with exogenous prices, with an optimal ﬁnite number
of products. The results show that endogenous pricing is a further
force toward an optimal smaller number of products if  is not also
increasing in n.
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
1342 Management Science 55(8), pp. 1338–1346, ©2009 INFORMS
Before consumers decide whether to search, the
ﬁrm chooses the number of products to sell, n, such
that it maximizes its proﬁt and foresees that the con
sumers are going to choose whether to search depend
ing on n and the ﬁrm’s optimal pricing strategy.
To emphasize again the timing of actions, the ﬁrm
chooses simultaneously the number of products and
their locations and prices; prior to searching, the con
sumers only observe the number of products; after
searching, the consumers are also able to observe the
product locations and prices.
3. Consumer and Firm Behavior
In this section, we consider the consumer behavior of
whether to search and the optimal decision of the ﬁrm
with respect to prices and the number of products.
3.1. Consumer Behavior
Consider the expected surplus of consumers after
they incur the search costs. Given the number of prod
ucts offered, n, consumers are able to anticipate the
equilibrium price charged for each product and to see
that in equilibrium the products are equally spaced
around the circle.
11
If the consumer’s valuation . is sufﬁciently greater
than the price p, the consumer will always purchase
a product. As there are n products, the furthest that
a product can be from the ideal product of the con
sumer is 1¡(2n), with an expected distance from the
best product of 1¡(4n). The expected surplus obtained
by the consumer is then . −p −¡(4n).
However, if the consumer’s valuation . is not much
greater than the price p, the consumer only buys if
his or her ideal product is sufﬁciently close to one
of the products. This will occur if (. −p)¡  1¡(2n).
In this case, the probability of a consumer purchas
ing the product after evaluating the product line is
2n((. −p)¡) and the expected surplus, given that the
consumer purchases a product, is . − p − ((. − p)¡
(2)) = (. − p)¡2. The expected surplus after search
ing unconditionally on the location of the consumer’s
ideal product is then ((. −p)
2
¡)n.
From this, if . ≥p +¡(2n), the marginal consumer .
is determined by . −p −¡(4n) =, and all consumers
that search buy a product. Yet if .  p + ¡(2n), the
marginal consumer . is determinedby (( .−p)
2
¡)n=,
11
As noted below, the equilibrium price is the same across prod
ucts, and the optimal locations are equally spaced around the cir
cle. Even though the equilibrium price is the same across products,
consumers only ﬁnd out about the actual prices after visiting the
store and paying the search costs . Note that if consumers could
buy at random without checking prices, then their always buying
at random cannot be an equilibrium, because otherwise they could
be charged a very high price for one product. Note also that in
such a setting, checking the price of only one product would not
guarantee that the prices of the other products would be the same.
and some of the consumers that search after inspect
ing the products decide not to buy any product.
Depending on the ﬁrm behavior, the market can be in
either case.
3.2. Firm Pricing Decision and Market Outcome
Given the Product Line Length
Consider now the ﬁrm’s pricing decision. Note that
given that consumers have searched, the best thing
that the ﬁrm can do to extract the most surplus from
the consumers who have searched is to offer equally
spaced products at the same price.
If the optimal price will be such that . ≥p +¡(2n),
such that all consumers after searching choose to buy
a product, then the optimal price is the one that makes
the marginal consumers at the midpoint between two
products to be indifferent between buying and not
buying a product, p = . −¡(2n). Then, independent
of which case we are in (either all consumers that
search buy or only some consumers that search buy),
we have that the optimal price satisﬁes p ≥ .−¡(2n).
Given this condition, the total demand given the
price p, the number of products n, and the thresh
old ., D(p, n, .), is
D(p,n, .) =
.
p+¡(2n)
dí (.)+
p+¡(2n)
.
2n
.−p

dí (.), (1)
where the ﬁrst integral represents the demand of the
consumers with a sufﬁciently high valuation such that
they always buy after searching, and the second inte
gral represents the demand of the consumers who
only buy after searching if their preferences end up
sufﬁciently close to one of the available products, and
if there is any positive mass of such consumers. Note
that the second integral is equal to zero if all con
sumers that search decide to buy, as p = . −¡(2n) in
that case.
The optimal price—given the number of products
n and the marginal consumer’s valuation .—results
from the maximization of the ﬁrm’s proﬁt pD(p, n, .),
which results in the ﬁrstorder condition
1 −í
p +

2n
−p
4n

í
p +

2n
−í ( .)
+
p+¡(2n)
.
2n

.dí (.) =0. (2)
Note that the lefthand side of (2) is strictly positive
when p = . − ¡(2n), which means that the optimal
price is p > .−¡(2n); that is, the case in which all con
sumers that search end up buying a product is never
an equilibrium. The equilibrium price p and marginal
consumer’s valuation . is then determined by jointly
solving (2) and (( . −p)
2
¡)n = . Totally differentiat
ing these two equations with respect to the price p,
the marginal consumer valuation ., and the number
of products offered n one obtains the following result
(the proof is presented in the appendix).
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
Management Science 55(8), pp. 1338–1346, ©2009 INFORMS 1343
Proposition 1. When the number of products offered,
n, increases, the equilibrium price p increases by more than
the marginal consumer valuation ..
As the number of products increases, every con
sumer knows two things. First, once he incurs the
ﬁxed evaluation cost , he is more likely to ﬁnd a
product that ﬁts his taste. The ﬁrm knows this. This
would then lead to a better matching of consumers to
their favorite products, so consumers with very sim
ilar taste and willingness to pay will be matched to
any given product. Second, because the ﬁrm antic
ipates this, it can charge a higher price to extract
the consumer’s surplus. Foreseeing this, consumers
become less willing to purchase, i.e., the valuation
of the marginal consumer that searches, ., is likely
to be greater. Note that the price increases because
of the increase in both the number of products and
the marginal valuation of the consumers who search.
In contrast, the marginal valuation increases with the
price increase but decreases with the increase in the
number of products. Therefore, the effect on price of
an increase in the number of products is greater than
the effect on the marginal valuation.
Note also that for this equilibrium to exist we have
p + ¡(2n) > . = p +
¡n, which reduces to  
¡(4n). If  ≥ ¡(4n), the equilibrium is for no con
sumer to search, as consumers know that once they
search, the price they face will be so high that their
ex ante surplus would be negative. In other words,
if the ﬁrm chooses a number of products n that is
too high, then no consumer will search or buy. Con
sider the extreme case with an inﬁnite number of
products. Then all the consumer surplus would be
extracted, but because consumers know this, they will
not search. Then, as argued below, a ﬁrm will choose
optimally not to offer as many products.
To illustrate the results above for the case  
¡(4n), consider the example where . is distributed
uniformly in í0, .{. In that case (2) becomes . −p −
¡(2n)−(n¡)(¡(2n)−
¡n)(2p−¡(2n)−
¡n) =0,
generating an equilibrium price
p =
1
2
. −¡(4n) −
1 −
n¡
.
It is straightforward to check that the equilibrium
price is increasing in n and that it increases faster
than the equilibrium .. Figure 1 shows how . and the
price p vary with the number of products for . = 4,
 =2, and  =0.08. Note that for this example if n ≥7,
then  >¡(4n) and no consumer chooses to search.
Using the proof of Proposition 1,
12
one can also see
that when the search costs  increase, ﬁxing the num
ber of products n, the equilibrium price p, and val
uation of the marginal consumer . increase. As the
12
The slope d .¡dp of the condition . =p +
¡n is not as steep as
the same slope for (2) at the equilibrium.
Figure 1 Evolution of Equilibrium Valuation t of Marginal Consumers
and Price p as a Function of the Number of Products r, for t
Uniformly Distributed on 0, 4, t =2, and k =0.08
2
3
4
0 1 2 3 4 5 6 7
v
,
p
v
p
n
ˆ
ˆ
search costs increase, fewer consumers search and .
is higher, which leads to higher equilibrium prices.
3.3. Optimal Number of Products
Let us now consider the optimal decision by the ﬁrm
on the number of products to offer. We already noted
that if the number of products offered is too high,
such that  > ¡(4n), then no consumer searches and
the ﬁrm ends up with no demand. One can then
obtain that if the search costs  >0, then the optimal
number of products is ﬁnite. We state this result in
the following proposition (the proof is presented in
the appendix).
Proposition 2. Suppose that the search costs are
strictly greater than zero,  >0. Then the optimal number
of products is ﬁnite.
Consider now the question of what the optimal
number of products offered, n, is. Denoting p(n) as
the equilibrium price per product, given that n prod
ucts are offered, and .(n) the equilibrium valuation of
the marginal consumers searching, we can write the
problem of the ﬁrm as
max
n
p(n)D(p(n), n, .(n)), (3)
where D(p(n), n, .(n)) is directly obtained from (1).
By the envelope theorem, the derivative with respect
to n of (3) only needs to take into account the direct
effect of n on proﬁts and the effect of n through
the equilibrium .. The ﬁrstorder condition determin
ing n is then
p(n)
oD
on
+p(n)
oD
o .
d .
dn
=0. (4)
It is immediately seen that oD¡on >0. That is, keep
ing prices and the valuation of the marginal con
sumers ﬁxed, an increase in the number of products
increases demand. Furthermore, oD¡o .  0, keeping
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
1344 Management Science 55(8), pp. 1338–1346, ©2009 INFORMS
prices and the number of products ﬁxed, an increase
in the valuation of the marginal consumers reduces
demand. Then, from (4), we can obtain that close
to the optimum d .¡dn > 0, as the number of prod
ucts offered increases, the number of consumers that
search falls. By Proposition 1, we also know that, close
to the optimum, as the number of products offered
increases, the equilibrium price also increases. We
state these results in the following lemma.
Lemma 1. Given the optimal number of products n
∗
,
there exists a o > 0 such that for n ∈ ín
∗
− o, n
∗
+ o{,
as the number n of products offered increases, then both
the valuation of the marginal consumers searching and the
equilibrium price increases.
This result illustrates the effect of strategic pricing
on the optimal decision of the ﬁrms and on the deci
sion of consumers not to search—the choice overload
effect. That is, close to the optimum, as the num
ber of products increases, fewer consumers choose to
search (and ultimately do not purchase the product),
as the high prices that they fear may not allow them
to recoup the search costs incurred. This also implies
that close to the optimum increasing the number of
products decreases demand.
13
It is illustrative to investigate what happens when
the search costs  go to zero. In that case, for a ﬁxed n,
we have . converging to the price p and oD¡o . con
verging to zero, which from (4) implies that the opti
mal number of products offered converges to inﬁnity.
The condition on the equilibrium price (2) converges
then to 1 −í (p) −p1 (p) =0, which is the typical con
dition for monopoly pricing under no search costs.
One can then obtain the following result on the
effect of the search costs  on the optimal number of
products:
Proposition 3. For any search cost  there is a lower
search cost 
such that the optimal number of products
under search cost  is lower than under search cost 
. If the
probability distribution of the valuation . is uniform, then
the optimal number of products is decreasing in the search
costs  for all .
Figure 2 illustrates the optimal number of prod
ucts n as a function of the search costs  when . is dis
tributed uniformly on í0, 4{ and  =2. As noted in the
proposition, in general, one can obtain the result that,
for some region of the search costs, when search costs
increase, the optimal number of products decreases.
The intuition is that when the search costs increase,
the ﬁrm has to offer a greater surplus for consumers
13
However, note that increasing the number of products may
increase demand if the number of products is not close to the
optimum. To see this, consider, for example, the case of . uniformly
distributed,  close to zero, and n =1.
Figure 2 Optimal Number of Products r as a Function of Search
Costs k, for t Uniformly Distributed on 0, 4 and t =2
0
2
4
6
8
10
12
14
16
18
20
0 0.01 0.02 0.03 0.04 0.05
n
k
to be willing to search, and it can achieve that by
reducing the number of products offered, as a lower
number of products generates lower prices.
It is also interesting to note that if ﬁrms were able
to observe prices before incurring search costs, then
the optimal number of products to offer would be
inﬁnity, given that price demand increases in the
number of products offered.
14
However, as noted
above, if consumers can observe price but are unable
to observe product quality prior to incurring the
search costs, then the optimal number of products
would continue to be ﬁnite, as a larger number of
products would allow a ﬁrm to extract too much con
sumer surplus by offering lower product quality.
4. Concluding Remarks
This paper investigates the role of product evalua
tion costs on consumer behavior and ﬁrm strategies
when prices are endogenous and the ﬁrm chooses
the product line length. This paper shows that if con
sumers have to invest resources to learn about the
prices charged, then there is a choice overload effect
in the sense that a greater number of products offered
leads to fewer consumers being willing to choose.
This effect results from the consumers being aware
that more products will lead the ﬁrm to charge higher
prices, as the products ﬁt consumer preferences better.
This effect then also generates the effect that the opti
mal number of products is ﬁnite even though there is
no cost per product offered. In sum, this paper shows
that with product evaluation costs, endogenous pric
ing leads to fewer products being offered.
14
If search costs are incurred per product evaluated, then the opti
mal number of products to offer would still be ﬁnite if prices
are observed prior to consumers incurring the search costs. The
endogenous pricing effects discussed here would still be present if
search costs are incurred per product evaluated.
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
Management Science 55(8), pp. 1338–1346, ©2009 INFORMS 1345
As noted above, the results extend to the case
in which product quality is also a choice variable by
the ﬁrms that consumers have to check, and con
sumers observe price without incurring any search
cost. The results also replicate when the consumers
who value the product category most are also the ones
who care more about product ﬁt. This paper considers
the monopoly case. It would be interesting to investi
gate what happens when there is competition. In such
a case, if ﬁrms have sufﬁcient market power, the same
effects may hold; it would be interesting to investigate
the effect of the degree of competition. Another inter
esting issue to investigate would be the implications
on price discrimination and vertical differentiation of
product evaluation costs. The analysis above consid
ers only horizontal differentiation, but similar model
features may lead themselves to study vertical prod
uct differentiation.
Acknowledgments
Comments from the department editor, associate editor,
two anonymous reviewers, Emir Kamenica, Dmitri Kuksov,
Steven Shugan, and seminar participants at the University
of Texas at Dallas, Washington University, St. Louis, and the
Marketing in Israel conference are gratefully appreciated.
An earlier version of this paper (VillasBoas 2008) circu
lated under the title “Information Overload with Endoge
nous Pricing.”
Appendix
Proof of Proposition 1. Note ﬁrst that (2) implies that
p+¡(2n)
.
(2p−.) dí (.) >0, which implies 2p > .. Differentiat
ing (2) with respect to ., one obtains x ≡(2n¡)1 ( .)(2p − .),
which is positive as 2p > .. Adding x to the derivative of (2)
with respect to p one obtains
n ≡ −1
p +

2n
−
4n

í
p +

2n
−í ( .)
+
2n

1 ( .)(2p − .) −
2n

1
p +

2n
p −

2n
.
Note that for p = . −¡(2n) the condition (2) yields . = .,
which is greater than the . from the condition . =p+
¡n,
as ¡(2n) >
¡n, given that in equilibrium we need to
have p +¡(2n) > .. That implies that the slope d .¡dp of the
condition . = p +
¡n is less steep than the same slope
for (2) at the equilibrium, which implies n 0.
Totally differentiating the condition . = p +
¡n and
(2) with respect to ., p, and n, one obtains d .¡dn =
(z − u(n −x))¡n and dp¡dn = (z − ux)¡n, where u =
1
2
(¡n
2
)(¡n)
1¡2
and z is minus the derivative of (2)
with respect to n. One can then obtain dp¡dn − d .¡dn =
u(n −2x)¡n >0.
Proof of Proposition 2. Suppose the ﬁrm offers an
inﬁnite number of products, and the equilibrium has con
sumers with . ≥ . with . ≤ . incurring search costs . Note
that . ≥argmaxpí1−í (p){, as the ﬁrm would never charge
a price below argmaxpí1−í (p){. But then as each consumer
has his ideal product available, the ﬁrm will never charge a
price below .. But then a consumer with a valuation equal
to . would not incur the search costs , as after incurring
the search costs the consumer would get zero surplus, for
a total negative surplus. Then, with an inﬁnite number of
products, no consumer will purchase any product, and the
ﬁrm would end up with zero proﬁts. However, as noted
in the text, if n  ¡(4), ﬁnite, the ﬁrm gets strictly posi
tive demand and proﬁt. Therefore, the optimal number of
products is ﬁnite if positive proﬁts can be obtained.
Proof of Proposition 3. Suppose search costs are .
Then we know that the optimal number of products is
strictly below ¡(4). For search costs 
(below) converging
to zero we know that the optimal number of products con
verges to inﬁnity. Then the ﬁrst statement in the proposition
follows. Consider now the case where . is uniformly dis
tributed. For this case one can obtain .(oD¡on) = ¡(4n
2
) −
¡n, .(oD¡ .) =−2
n¡, and d .¡dn =
√
(((2
√
 −3
√
n) +
4n
√
n( .−)−8
√
n(
√
 −
√
n)
2
)¡(16n
2
(
√
 −
√
n)
2
)). Using
(4) one can then obtain the condition for the optimal n as
2
2
−6.
√
 +5.
2
−4 .n.
2
+4.
4
=0, (5)
where . is deﬁned as . ≡
√
n. Using (5) to write n as a
function of ., and taking the derivative of n with respect to
. one obtains dn¡d.  0 as the condition n  ¡(4) trans
lates into . 
√
¡2. As d.
2
¡dn = +n(d¡dn) we have that
dn¡d  0, a higher search cost leads to a lower optimal
number of products.
References
Beach, L. R., T. R. Mitchell. 1978. A contingency model for the selec
tion of decision strategies. Acad. Management Rev. 3(3) 439–449.
Bénabou, R., J. Tirole. 2004. Willpower and personal rules. J. Political
Econom. 112(4) 848–886.
Bettman, J. R. 1979. An Information Processing Theory of Consumer
Choice. AddisonWesley, Reading, MA.
Boatwright, P., J. C. Nunes. 2001. Reducing assortment: An
attributebased approach. J. Marketing 65(3) 50–63.
Dhar, R., I. Simonson. 2003. The effect of forced choice on choice.
J. Marketing Res. 40(2) 146–160.
Diamond, P. 1971. A model of price adjustment. J. Econom. Theory
3(2) 156–168.
Dickson, P. R., A. G. Sawyer. 1990. The price knowledge and search
of supermarket shoppers. J. Marketing 54(3) 42–53.
Fudenberg, D., D. K. Levine. 2006. A dualself model of impulse
control. Amer. Econom. Rev. 96(5) 1449–1476.
Gourville, J. T., D. Soman. 2005. Overchoice and assortment type:
When and why variety backﬁres. Marketing Sci. 24(3) 382–395.
Gul, F., W. Pesendorfer. 2001. Temptation and selfcontrol.
Econometrica 69(6) 1403–1436.
Guo, L. 2006. Consumption ﬂexibility, product conﬁguration, and
market competition. Marketing Sci. 25(2) 116–130.
Hauser, J. R., B. Wernerfelt. 1990. An evaluation cost model of con
sideration sets. J. Consumer Res. 16(4) 393–408.
Huffman, C., B. E. Kahn. 1998. Variety for sale: Mass customization
or mass confusion? J. Retailing 74(4) 491–513.
Irons, B., C. Hepburn. 2007. Regret theory and the tyranny of
choice. Econom. Record 83(261) 191–203.
Iyengar, S. S., E. Kamenica. 2007. Choice proliferation, simplicity
seeking, and asset allocation. Working paper, Columbia Uni
versity, New York.
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs
1346 Management Science 55(8), pp. 1338–1346, ©2009 INFORMS
Iyengar, S. S., M. R. Lepper. 2000. When choice is demotivating:
Can one desire too much of a good thing? J. Personality Soc.
Psych. 79(6) 995–1006.
Jacoby, J. 1977. Information load and decision quality: Some con
tested issues. J. Marketing Res. 14(4) 569–573.
Jacoby, J., J. C. Olson. 1977. Consumer response to price:
An attitudinal, information processing perspective. Y. Wind,
B. Greenberg, eds. Moving Ahead with Attitude Research.
American Marketing Association, Chicago, 73–86.
Kamenica, E. 2007. Contextual inference in markets: On the infor
mational content of product lines. Working paper, University
of Chicago, Chicago.
Kamenica, E. 2008. Contextual inference in markets: On the infor
mational content of product lines. Amer. Econom. Rev. 98(5)
2127–2149.
Keller, K. L., R. Staelin. 1987. Effects of quality and quantity of
information on decision effectiveness. J. Consumer Res. 14(2)
200–213.
Kuksov, D. 2004. Buyer search costs and endogenous product
design. Marketing Sci. 23(4) 490–499.
Kuksov, D., J. M. VillasBoas. 2006. When more alternatives lead to
less choice. Working paper, University of California, Berkeley,
Berkeley.
Lal, R., C. Matutes. 1994. Retail pricing and advertising strategies.
J. Bus. 67(3) 345–370.
Loomes, G., R. Sugden. 1987. Some implications of a more general
form of regret theory. J. Econom. Theory 41(2) 270–287.
Lurie, N. H. 2004. Decision making in informationrich environ
ments: The role of information structure. J. Consumer Res. 30(4)
473–486.
Mick, G. L., S. M. Broniarczyk, J. Haidt. 2004. Choose, choose,
choose, choose, choose, choose, choose: Emerging and prospec
tive research on the deleterious effects of living in consumer
hyperchoice. J. Bus. Ethics 52(2) 207–211.
Miller, G. A. 1956. The magical number seven, plus or minus two:
Some limits of our capacity for processing information. Psych.
Rev. 63(2) 81–97.
Norwood, F. B. 2006. Less choice is better, sometimes. J. Agricultural
Food Indust. Organ. 4(1) Article 3.
Payne, J. W., J. R. Bettman, E. J. Johnson. 1988. Adaptive strategy
selection in decision making. J. Experiment. Psych.: Learn., Mem
ory, Cognition 14(3) 534–552.
Rao, R. C., N. Syam. 2001. Equilibrium price communication and
unadvertised specials by competing supermarkets. Marketing
Sci. 20(1) 61–81.
Sarver, T. 2008. Anticipating regret: Why fewer options may be
better. Econometrica 76(2) 263–305.
Shugan, S. M. 1980. The cost of thinking. J. Consumer Res. 7(2)
99–111.
Simester, D. 1995. Signalling price image using advertised prices.
Marketing Sci. 14(2) 166–188.
Simon, H. A. 1955. A behavioral model of rational choice. Quart. J.
Econom. 69(1) 99–118.
Van Zandt, T. 2004. Information overload in a network of targeted
communication. RAND J. Econom. 35(3) 542–560.
VillasBoas, J. M. 2004. Communication strategies and product line
design. Marketing Sci. 23(3) 304–316.
VillasBoas, J. M. 2008. Information overload with endogenous
pricing. Working paper, University of California, Berkeley,
Berkeley.
Wathieu, L., M. Bertini. 2007. Price as a stimulus to think: The case
of willful overpricing. Marketing Sci. 26(1) 118–129.
Wernerfelt, B. 1994. Selling formats for search goods. Marketing Sci.
13(3) 298–308.
Zhang, J. 2009. The sound of silence: Observational learning in the
U.S. kidney market. Marketing Sci., ePub ahead of print July 23,
http://mktsci.journal.informs.org/cgi/content/abstract/mksc.
1090.0500v1.
I
N
F
O
R
M
S
h
o
l
d
s
c
o
p
y
r
i
g
h
t
t
o
t
h
i
s
a
r
t
i
c
l
e
a
n
d
d
i
s
t
r
i
b
u
t
e
d
t
h
i
s
c
o
p
y
a
s
a
c
o
u
r
t
e
s
y
t
o
t
h
e
a
u
t
h
o
r
(
s
)
.
A
d
d
i
t
i
o
n
a
l
i
n
f
o
r
m
a
t
i
o
n
,
i
n
c
l
u
d
i
n
g
r
i
g
h
t
s
a
n
d
p
e
r
m
i
s
s
i
o
n
p
o
l
i
c
i
e
s
,
i
s
a
v
a
i
l
a
b
l
e
a
t
h
t
t
p
:
/
/
j
o
u
r
n
a
l
s
.
i
n
f
o
r
m
s
.
o
r
g
/
.
as only the better alternatives can generate some signiﬁcant surplus. INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). By offering a smaller number of products. and prices are observed prior to the consumers incurring the evaluation costs. For example. In many market situations. and the consumer only purchases a product if the product offers a positive surplus (excluding the evaluation costs.informs.VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8). even if the product attributes or prices are displayed or advertised. in order to isolate the essence of the effect of endogenous pricing. The ﬁrm can then beneﬁt from reducing the surplus of the consumers (excluding the evaluation costs) by increasing prices or reducing product quality. In the formulation below. With a smaller number of products.2 For example. if these can be checked only after the products are evaluated. Greater service in these cases may mean the ability to respond to the speciﬁc consumer demands. Processing the information of the attributes of each product and the prices being charged may be costly or take time. © 2009 INFORMS 1339 products. as the existence of evaluation costs makes the consumers choose not to buy (and not incur the evaluation costs) when faced with too many alternatives. or when a consumer sees the total list of products available on a webpage. consumers may have ﬁrst a sense of the number of product available. For example. however. endogenous pricing leads to a small number of conceptual model of price encoding and knowledge. Supermarkets that offer a large variety in a certain category may be able to charge higher prices because they provide better productconsumer ﬁt. Iyengar and Kamenica 2007).org/. they are sunk costs. the consumers who evaluate the products are the ones who have a higher general valuation for the category. Additional information. generating higher ex post surplus for the consumers who search. This then gives less incentive for consumers to evaluate the product (as they know that they will have a reduced surplus after evaluation). restaurants that offer greater variety in service may be able to charge higher prices for their speciﬁc dishes. If consumers at the time of purchase are unsure of their preferences at the time of consumption (Guo 2006). and some consumers who evaluate the products end up not buying any product. after observing the number of products available. and the ﬁrm ends up with less demand. which may yield lower consumer surplus. 3 See Dickson and Sawyer (1990) on the extent to which consumers are uninformed about prices after purchase decisions and on a evaluation costs are incurred. pp. The same results (available upon request) can be obtained when ﬁrms also make decisions on the quality of each product. In some markets where the information on prices is more costly to obtain or is less advertised. is available at http://journals. . That is. the interaction between ﬁrm pricing and quality decisions and evaluation costs is a force toward a smaller number of products. Then. In this setting. consumers realize that all consumers will ﬁnd a product that provides a good ﬁt (low variance of consumers valuations around best product ﬁt). a consumer has to decide whether it is worth incurring evaluation costs to ﬁnd out the value of each product for the consumer. which is only found out after the consumer evaluation. Sellers with greater variety may be able to charge more speciﬁc (higher) prices. greater service is the ability to better respond to the speciﬁc demands of each customer (greater variety of services provided). or hairdressers.3 But once the 2 Note. which means that consumers have to ﬁnd that product. which can allow the ﬁrm to extract more consumer surplus. which could be a force toward lower prices.g. that this effect holds when keeping other factors ﬁxed. they may beneﬁt less from searching and processing information. hotels. One can obtain similar results when the general consumer valuation for a category is positively correlated with how much a consumer cares about product ﬁt. The optimal number of products is decreasing in the evaluation costs and converges to inﬁnity when the evaluation costs go to zero. and information processing with respect to prices (see also Jacoby and Olson 1977). If the evaluation costs are high enough. as it may still ﬁnd it proﬁtable to attract consumers who do not have such a great ﬁt with any product (which is more likely with a smaller number of products). as they are sunk at the purchase stage). including rights and permission policies. evaluation costs are not allowed to vary with the number of products offered. sellers that offer more variety could have lower costs. 1338–1346. and the ﬁrm will extract the most it can from the surplus generated by that good ﬁt. Examples of some types of services where these effects may be present are highend restaurants. if there are economies of scope in variety. In some markets for services. The main results here may also suggest that ﬁrms that advertise more (and advertise their prices) may have more dense product lines. the ﬁrm can commit that it will offer some signiﬁcant consumer surplus.. Note that the effects of the evaluation costs presented here can be seen as choice overload effects (e. keeping other factors ﬁxed. the market can disappear. the ﬁrm lowers its prices to attract the consumers who do not have a good product ﬁt. we may expect that these effects are more important. This idea is present in markets where many alternatives are offered and consumers realize that ﬁnding the best alternative is important. This paper considers the pricing case and shows that there is an interior and ﬁnite optimal number of products. as the effects discussed here will not be present. given the speciﬁc needs satisﬁed. Consider a situation when consumers are able to observe ﬁrst the length of the product line (number of products) offered by a ﬁrm. In equilibrium. this could be the case when a consumer sees the length of shelf space provided in a supermarket or department store for a category.
Jacoby 1977 and the references listed there) and the idea that decision makers may only be able to process a limited amount of information (e. for example. where the length of product line may be more easily observed than prices or product quality. Kamenica (2008) considers the possibility that the information implicit in the product line may affect customers’ beliefs about how much they would enjoy particular products. such that prices cannot signal product characteristics. Kuksov and VillasBoas 2006) that under exogenous pricing. Iyengar and Lepper 2000) that even though lager choice sets may reduce choice. 1338–1346. reducing overall consumer search costs and leading more consumers to choose an alternative. Mick et al. each selling only one product. Kamenica (2007) also considers endogenous prices with large retail cost uncertainty (unknown to the consumers).. In addition to that literature. Simon 1955. if the evaluation costs are increasing in the number of product offered (or searched). Boatwright and Nunes (2001) show that reducing the assortment size in a grocery store can increase sales. Beach and Mitchell (1978) address the issue of decision makers eliminating options on the basis of deviation from their ideal point. This work has argued that consumers may feel overwhelmed or overloaded with too many alternatives. some experimental work has found that consumers are more likely to make a choice and purchase when confronted with a smaller choice set. Huffman and Kahn (1998) discuss the issue of variety choice creating confusion in preferences (see also.org/. there is a literature that looks at the opportunism of sellers when consumers have to incur sunk evaluation costs before they make a purchase. Putting the two results together.. consumers are attracted to sellers with greater choice sets. For example. here there is only one ﬁrm selling multiple products. In a market setting. This idea goes back to the information overload literature (see. endogenous pricing would lead to an even smaller optimal number of products offered. We know from previous work (e. Lal and Matutes (1994) and Wernerfelt (1994) discuss the possibility of advertised prices as a form of the seller commiting not to exploit the consumer sunk costs. Diamond (1971) considers search costs in consumers ﬁnding out about prices with competing sellers. is available at http://journals. in the particular contexts that are considered. including rights and permission policies. Some of the issues discussed in that paper can also be seen from the side of the ﬁrms through their communication strategies (VillasBoas 2004). 2004. As noted above.5 In this regard. and the ﬁrm can choose product attributes and the number of products it sells. Note also that there may be consumer processing costs for advertised prices. Bettman 1979). Hauser and Wernerfelt (1990) have argued that consumers may strategically limit their consideration sets (with search under ﬁxed sampling) to reduce evaluation costs. Kuksov and VillasBoas (2006) consider search costs in evaluating the ﬁt of each alternative and show that a strategic supplier of alternatives (spanning the alternatives through the product space) does better by offering a ﬁnite number of alternatives than an inﬁnite number.. this paper considers the possibility of ﬁrms choosing a product line. optimal products.6 Norwood (2006) considers freeentry price competition among ﬁxed products that are 5 Keller and Staelin (1987) argue that with regard to information. and if prices can be observed without processing or evaluation costs. Literature Review Related to this paper. This issue is further discussed in the model setup. the optimal number of products offered may be small. unobserved quality decisions generate the same problems (see discussion above).g. given some level of conﬁdence. pp. and exogenous pricing leads to an inﬁnite number of products.informs. Gourville and Soman 2005).1. this paper also notes that the product line speciﬁcation allows ex ante evaluation costs to generate information/choice overload effects. See also Kuksov (2004) for the effect of search costs on product design under competition. and each consumer having a downwardsloping demand. consumers might not have full information about search costs and might consider larger choice sets better given the information that they have. we would have that in a setting where the evaluation costs are increasing in the number of products offered.g. . 6 In a working paper version. 1988). 1. This paper will not address this later effect. Van Zandt (2004) considers competition where ﬁrms communicate about their products and consumers evaluate a limited and ﬁxed number of alternatives. There is also some evidence (e. Lurie 2004. Shugan (1980) considers the costs of thinking and provides a quantitative measure of that cost related to the number of comparisons necessary to make a decision. In contrast. which could depend on what is kept constant when choice sets are enlarged.. © 2009 INFORMS INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). Miller 1956) and mental 4 See also Simester (1995) and Rao and Syam (2001). he ﬁnds that there is too much communication in equilibrium. Additional information. The length of the product line commits the seller to the extent of consumer surplus that is provided. quality of information may help decision making.g. In relation to that literature. evaluation costs (e. whereas quantity of information may affect decision making negatively. Another important aspect that is not considered here involves the possibility of using attribute information to eliminate alternatives (Payne et al.4 and Wernerfelt also discusses the possibility of return policies and seller colocation. as a ﬁrm communicating about its product does not consider the negative externality on consumer information processing affecting the other ﬁrms.1340 VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8).g.
The remainder of this paper is organized as follows. a consumer searches if and only if Ed max v − td − p 0 n ≥ k. Search costs could also be different across consumers without affecting the main results presented here. If the product consumed is at a distance d of the consumer’s ideal product. For example. Note that if the search costs k are increasing in the number of products offered n. where Ed is the expected value operator over the distance d to the best product for that consumer. Additional information. with the cumulative distribution function F v (density f v ).g. That setup would complicate the analysis. is available at http://journals. Fudenberg and Levine 2006) and regret preferences (see. . Norwood includes an approximation to the consumer sequential evaluation process and ﬁnds that there is excessive entry in equilibrium. for example. we would then immediately obtain choice overload effects with exogenous prices. generating further incentives for consumers to invest in evaluating the purchase decision. 1338–1346. the existence of ﬁxed costs per product carried can generate a limited product line. which is enough to generate information overload effects with endogenous pricing. for example. Dhar and Simonson 2003). this paper focusses on the explanation based on simple evaluation costs. Note that this explanation could then be seen as a search cost explanation for no choice when many alternatives are presented. the consumers learn the location of all the products offered and their prices and the location of Wathieu and Bertini (2007) consider the effect of the price being close to the expected willingness to pay. Base Model Consider a ﬁrm deciding how many and which products at what prices to offer to a set of consumers.10 Consumers are heterogeneous on their general valuation v for a product in the category and on the location of their ideal product x in the circle. 8 For example. The ﬁrm chooses the prices of the different products (and their location) to maximize proﬁts given that consumers who have searched have v ≥ v. it may also be that a greater number of alternatives leads the decision maker to delay choice (and not choose when the choice set is ﬁrst presented) in order to gather more information on the choice problem (e.7 Note that the objective of this paper is not to argue that this explanation is the most compelling one for the existence of limited product lines.informs. under the assumption that only the most popular products are offered. Irons and Hepburn 2007. if and only if v − td − p ≥ 0. After incurring the search costs. © 2009 INFORMS 1341 INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). The marginal distribution of x is assumed uniform along the unit circle. we consider the effect of the interaction between the consumer evaluation costs and ﬁrm decision making about variables that affect the consumer surplus (price and quality) on the optimal number of products for a ﬁrm to offer. This new information could come at a lower cost than the cost of evaluating different alternatives at the present. vertically differentiated. and the price of each alternative. but the main messages of the results presented here should also extend to that setting. Section 3 presents the main results. Sarver 2008). there is going to be a threshold v such that a consumer searches if and only if v ≥ v. including rights and permission policies. their ideal product x. the consumer does not know d or p but may be able to infer both the distribution of d from the best product available and the price p from the number of products offered n. consumers know their general valuation v for the category and observe the number of alternatives n being offered by the ﬁrm. as the search costs are sunk at the purchase decision time. Note that the consumers foresee this behavior by the ﬁrm. therefore. In relation to other explanations for choice overload. As the expected value is increasing in v. takes into account the ﬁrm’s pricing strategy.. although it would not be central to the market forces presented here. Loomes and Sugden 1987. pp. the location of each of these alternatives on the circle. with an optimal ﬁnite number of products. the determination of v. one product per ﬁrm. 9 Consumers could also have search costs per additional product being evaluated. the valuation of the consumers indifferent between searching and not searching. a consumer chooses to buy the best product available. The model is presented in §2. 10 2. The results show that endogenous pricing is a further force toward an optimal smaller number of products if k is not also increasing in n. A consumer with valuation v decides to incur the search costs k if the expected surplus after search is greater than k. The ex post utility of a consumer with valuation v purchasing a product at price p at distance d of his ideal product is U = v − td − p − k Prior to incurring the search costs k.9 The ﬁrm decides the number of alternatives to offer n. which is at distance d. The distribution of v and x is assumed independent among consumers. Finally.org/. the consumer incurs a disutility td. After searching and incurring the search costs k. the paper presents an additional force toward a smaller number of products offered. Bénabou and Tirole 2004. Here. and the marginal distribution of v has positive density on the support 0 v . Before deciding whether to search. the discussion in Gul and Pesendorfer 2001.8 In relation to the papers above. where t is a parameter measuring the importance of the product ﬁt. consumers could gain some information about what other consumers have chosen (Zhang 2009). The products are located on a circle of unit length and have zero production costs (without loss of generality). and §4 concludes. 7 Note also that consumers could prefer smaller choice sets because of selfcontrol problems (see. That is.VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8).
This will occur if v − p /t < 1/ 2n . Consumer and Firm Behavior In this section. given that the consumer purchases a product. the marginal consumer v is determined by v − p 2 /t n=k. Note that if consumers could buy at random without checking prices. Note that given that consumers have searched. is v − p − t v − p / 2t = v − p /2.2. 1338–1346. Even though the equilibrium price is the same across products. the marginal consumer valuation v. if the consumer’s valuation v is not much greater than the price p. Note also that in such a setting. then the optimal price is the one that makes the marginal consumers at the midpoint between two products to be indifferent between buying and not buying a product. Given this condition. which results in the ﬁrstorder condition 4n t t −p F p+ −F v 1−F p+ 2n t 2n 2n v dF v = 0 (2) t v Note that the lefthand side of (2) is strictly positive when p = v − t/ 2n . Note that the second integral is equal to zero if all consumers that search decide to buy. consumers only ﬁnd out about the actual prices after visiting the store and paying the search costs k. the number of products n. As there are n products. To emphasize again the timing of actions. 3. which means that the optimal price is p > v −t/ 2n . The expected surplus obtained by the consumer is then v − p − t/ 4n . n. then their always buying at random cannot be an equilibrium. is D p n v = v p+t/ 2n 3. as p = v − t/ 2n in that case. because otherwise they could be charged a very high price for one product. including rights and permission policies. the ﬁrm chooses simultaneously the number of products and their locations and prices. after searching. + p+t/ 2n . In this case. the consumers only observe the number of products. n. the consumer will always purchase a product. such that all consumers after searching choose to buy a product. we consider the consumer behavior of whether to search and the optimal decision of the ﬁrm with respect to prices and the number of products. p = v − t/ 2n . and the threshold v. Consumer Behavior Consider the expected surplus of consumers after they incur the search costs. the case in which all consumers that search end up buying a product is never an equilibrium. the consumers are also able to observe the product locations and prices. If the optimal price will be such that v ≥ p + t/ 2n .org/. The equilibrium price p and marginal consumer’s valuation v is then determined by jointly solving (2) and v − p 2 /t n = k. prior to searching. The optimal price—given the number of products n and the marginal consumer’s valuation v—results from the maximization of the ﬁrm’s proﬁt pD p n v . and the second integral represents the demand of the consumers who only buy after searching if their preferences end up sufﬁciently close to one of the available products. The expected surplus after searching unconditionally on the location of the consumer’s ideal product is then v − p 2 /t n. that is. Totally differentiating these two equations with respect to the price p. 3. the ﬁrm chooses the number of products to sell. with an expected distance from the best product of 1/ 4n .1342 VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8). and some of the consumers that search after inspecting the products decide not to buy any product. the equilibrium price is the same across products. we have that the optimal price satisﬁes p ≥ v − t/ 2n . the consumer only buys if his or her ideal product is sufﬁciently close to one of the products.1. 11 As noted below. such that it maximizes its proﬁt and foresees that the consumers are going to choose whether to search depending on n and the ﬁrm’s optimal pricing strategy. is available at http://journals. independent of which case we are in (either all consumers that search buy or only some consumers that search buy). and all consumers that search buy a product. the total demand given the price p. and the number of products offered n one obtains the following result (the proof is presented in the appendix). if v ≥ p + t/ 2n . the furthest that a product can be from the ideal product of the consumer is 1/ 2n . Additional information. checking the price of only one product would not guarantee that the prices of the other products would be the same. consumers are able to anticipate the equilibrium price charged for each product and to see that in equilibrium the products are equally spaced around the circle. However. Then. the probability of a consumer purchasing the product after evaluating the product line is 2n v − p /t and the expected surplus. From this.11 If the consumer’s valuation v is sufﬁciently greater than the price p. dF v + p+t/ 2n 2n v v −p dF v t (1) where the ﬁrst integral represents the demand of the consumers with a sufﬁciently high valuation such that they always buy after searching. Depending on the ﬁrm behavior. and the optimal locations are equally spaced around the circle. the marginal consumer v is determined by v − p − t/ 4n = k. and if there is any positive mass of such consumers.informs. D p n v . Firm Pricing Decision and Market Outcome Given the Product Line Length Consider now the ﬁrm’s pricing decision. the best thing that the ﬁrm can do to extract the most surplus from the consumers who have searched is to offer equally spaced products at the same price. Given the number of products offered. Yet if v < p + t/ 2n . pp. the market can be in either case. © 2009 INFORMS INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). Before consumers decide whether to search.
Denoting p n as the equilibrium price per product. Consider the extreme case with an inﬁnite number of products. every consumer knows two things. which reduces to k < t/ 4n . Therefore. once he incurs the ﬁxed evaluation cost k. keeping . As the 12 The slope d v/dp of the condition v = p + kt/n is not as steep as the same slope for (2) at the equilibrium. i. increases. the derivative with respect to n of (3) only needs to take into account the direct effect of n on proﬁts and the effect of n through the equilibrium v. then k > t/ 4n and no consumer chooses to search. We state this result in the following proposition (the proof is presented in the appendix). as consumers know that once they search. When the number of products offered. then no consumer searches and the ﬁrm ends up with no demand. and k = 0 08. the valuation of the marginal consumer that searches. the equilibrium price p. Figure 1 shows how v and the price p vary with the number of products for v = 4. such that k > t/ 4n . To illustrate the results above for the case k < t/ 4n . consumers become less willing to purchase. If k ≥ t/ 4n . then no consumer will search or buy. n. 3. Using the proof of Proposition 1. it can charge a higher price to extract the consumer’s surplus. if the ﬁrm chooses a number of products n that is too high. where D p n n v n is directly obtained from (1). Then all the consumer surplus would be extracted. consider the example where v is distributed uniformly in 0 v . the equilibrium price p increases by more than the marginal consumer valuation v As the number of products increases. the marginal valuation increases with the price increase but decreases with the increase in the number of products. so consumers with very similar taste and willingness to pay will be matched to any given product. Note that for this example if n ≥ 7. In that case (2) becomes v − p − t/ 2n − n/t t/ 2n − kt/n 2p −t/ 2n − kt/n = 0. ﬁxing the number of products n. n. is available at http://journals.VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8). is likely to be greater. First. Note also that for this equilibrium to exist we have p + t/ 2n > v = p + kt/n. for v Uniformly Distributed on 0 4 . a ﬁrm will choose optimally not to offer as many products. We already noted that if the number of products offered is too high. the price they face will be so high that their ex ante surplus would be negative.e. Consider now the question of what the optimal number of products offered. because the ﬁrm anticipates this. an increase in the number of products increases demand. the effect on price of an increase in the number of products is greater than the effect on the marginal valuation. D/ v < 0. Then the optimal number of products is ﬁnite. fewer consumers search and v is higher. keeping prices and the valuation of the marginal consumers ﬁxed.org/. This would then lead to a better matching of consumers to their favorite products. The ﬁrm knows this. t = 2. Proposition 2. © 2009 INFORMS 1343 Evolution of Equilibrium Valuation v of Marginal Consumers and Price p as a Function of the Number of Products n. v. By the envelope theorem. The ﬁrstorder condition determining n is then pn D dv D +p n =0 n v dn (4) It is immediately seen that D/ n > 0. including rights and permission policies. Note that the price increases because of the increase in both the number of products and the marginal valuation of the consumers who search. he is more likely to ﬁnd a product that ﬁts his taste. t = 2. is. In contrast. generating an equilibrium price p= 1 v − t/ 4n − k 2 1 − nk/t Figure 1 4 INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s).3. k > 0. pp. they will not search. 1338–1346. as argued below. Suppose that the search costs are strictly greater than zero. Foreseeing this. Additional information. v. which leads to higher equilibrium prices. Second. Optimal Number of Products Let us now consider the optimal decision by the ﬁrm on the number of products to offer. p ˆ 3 v ˆ p 2 0 1 2 3 4 5 6 7 n search costs increase. Furthermore.informs. In other words. and k = 0 08 Proposition 1.. That is. we can write the problem of the ﬁrm as max p n D p n n v n n (3) It is straightforward to check that the equilibrium price is increasing in n and that it increases faster than the equilibrium v. One can then obtain that if the search costs k > 0. given that n products are offered. but because consumers know this. Then.12 one can also see that when the search costs k increase. and valuation of the marginal consumer v increase. and v n the equilibrium valuation of the marginal consumers searching. then the optimal number of products is ﬁnite. the equilibrium is for no consumer to search.
the optimal number of products decreases. Figure 2 illustrates the optimal number of products n as a function of the search costs k when v is distributed uniformly on 0 4 and t = 2.02 0. 1338–1346. endogenous pricing leads to fewer products being offered. note that increasing the number of products may increase demand if the number of products is not close to the optimum. then the optimal number of products to offer would still be ﬁnite if prices are observed prior to consumers incurring the search costs. Then.05 k to be willing to search. consider. k close to zero. one can obtain the result that. In sum. and n = 1. pp. This effect then also generates the effect that the optimal number of products is ﬁnite even though there is no cost per product offered. the case of v uniformly distributed. which from (4) implies that the optimal number of products offered converges to inﬁnity. Lemma 1. as a larger number of products would allow a ﬁrm to extract too much consumer surplus by offering lower product quality. the ﬁrm has to offer a greater surplus for consumers 13 Figure 2 20 18 16 14 12 Optimal Number of Products n as a Function of Search Costs k. there exists a > 0 such that for n ∈ n∗ − n∗ + . It is also interesting to note that if ﬁrms were able to observe prices before incurring search costs. the equilibrium price also increases.01 0. when search costs increase. if consumers can observe price but are unable to observe product quality prior to incurring the search costs. For any search cost k there is a lower search cost k such that the optimal number of products under search cost k is lower than under search cost k . Given the optimal number of products n∗ . as the number n of products offered increases. as noted above. then the optimal number of products is decreasing in the search costs k for all k. given that price demand increases in the number of products offered. in general. this paper shows that with product evaluation costs. and it can achieve that by reducing the number of products offered. close to the optimum. an increase in the valuation of the marginal consumers reduces demand. for v Uniformly Distributed on 0 4 and t = 2 n 10 8 6 4 2 0 0 0. This paper shows that if consumers have to invest resources to learn about the prices charged. is available at http://journals. 4. for a ﬁxed n. © 2009 INFORMS INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). as the high prices that they fear may not allow them to recoup the search costs incurred. then both the valuation of the marginal consumers searching and the equilibrium price increases. The condition on the equilibrium price (2) converges then to 1 − F p − pf p = 0. close to the optimum. 14 However. One can then obtain the following result on the effect of the search costs k on the optimal number of products: Proposition 3. as a lower number of products generates lower prices. To see this. then the optimal number of products would continue to be ﬁnite.informs.14 However. as the number of products offered increases. The intuition is that when the search costs increase. as the number of products offered increases. for some region of the search costs. We state these results in the following lemma. In that case.org/.13 It is illustrative to investigate what happens when the search costs k go to zero.03 0. then there is a choice overload effect in the sense that a greater number of products offered leads to fewer consumers being willing to choose. we have v converging to the price p and D/ v converging to zero. By Proposition 1. including rights and permission policies. as the products ﬁt consumer preferences better.04 0. The endogenous pricing effects discussed here would still be present if search costs are incurred per product evaluated.1344 VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8). fewer consumers choose to search (and ultimately do not purchase the product). prices and the number of products ﬁxed. That is. . from (4). for example. If search costs are incurred per product evaluated. we can obtain that close to the optimum d v/dn > 0. Additional information. then the optimal number of products to offer would be inﬁnity. This result illustrates the effect of strategic pricing on the optimal decision of the ﬁrms and on the decision of consumers not to search—the choice overload effect. we also know that. This also implies that close to the optimum increasing the number of products decreases demand. If the probability distribution of the valuation v is uniform. as the number of products increases. Concluding Remarks This paper investigates the role of product evaluation costs on consumer behavior and ﬁrm strategies when prices are endogenous and the ﬁrm chooses the product line length. This effect results from the consumers being aware that more products will lead the ﬁrm to charge higher prices. As noted in the proposition. the number of consumers that search falls. which is the typical condition for monopoly pricing under no search costs.
Management Rev. Then the ﬁrst statement in the proposition follows. Levine. 25(2) 116–130. Temptation and selfcontrol.. Sawyer..informs.VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8). 40(2) 146–160. as after incurring the search costs the consumer would get zero surplus. Rev. 2001. Fudenberg. J. Additional information. An evaluation cost model of consideration sets. and asset allocation. That implies that the slope d v/dp of the condition v = p + kt/n is less steep than the same slope for (2) at the equilibrium. J. if ﬁrms have sufﬁcient market power. Note ﬁrst that (2) implies that 2p − v dF v > 0. Soman. 16(4) 393–408. In such a case. Tirole. Econom. for a total negative surplus. Mitchell. E. associate editor. Theory 3(2) 156–168. and seminar participants at the University of Texas at Dallas. 2007. no consumer will purchase any product. R. The results also replicate when the consumers who value the product category most are also the ones who care more about product ﬁt. J. Emir Kamenica. Nunes. An earlier version of this paper (VillasBoas 2008) circulated under the title “Information Overload with Endogenous Pricing. J. J. Regret theory and the tyranny of choice. Pesendorfer. Columbia University. © 2009 INFORMS 1345 INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). and d v/dn = √ t 2 t − 3 nk + t √ √ √ √ √ 4n nk v −k −8 nk t − nk 2 / 16n2 t − nk 2 . which is greater than the v from the condition v = p + kt/n. Bénabou. Dhar.org/. New York. Suppose the ﬁrm offers an inﬁnite number of products. Suppose search costs are k.. it would be interesting to investigate the effect of the degree of competition. with an inﬁnite number of products. the optimal number of products is ﬁnite if positive proﬁts can be obtained. P. Totally differentiating the condition v = p + kt/n and (2) with respect to v. R. R. and the Marketing in Israel conference are gratefully appreciated. An Information Processing Theory of Consumer Choice. D. Dmitri Kuksov. Irons. Huffman. K. Appendix Proof of Proposition 1. 2003. But then as each consumer has his ideal product available. B. Hauser. Wernerfelt.. as the ﬁrm would never charge a price below arg max p 1−F p . Marketing 54(3) 42–53. J. simplicity seeking. Reading. B. R. I. but similar model features may lead themselves to study vertical product differentiation. Louis. S. 24(3) 382–395. As noted above. Therefore. the same effects may hold. and the ﬁrm would end up with zero proﬁts.. Bettman. 1978. Overchoice and assortment type: When and why variety backﬁres. Willpower and personal rules. S. J. The analysis above considers only horizontal differentiation. Retailing 74(4) 491–513. Proof of Proposition 2. T. where a = 1 kt/n2 kt/n 1/2 and z is minus the derivative of (2) 2 with respect to n One can then obtain dp/dn − d v/dn = a y − 2x /y > 0. 2005. 1990. Guo. For this case one can obtain v D/ n = t/√ 2 − 4n √ √ k/n. if n < t/ 4k . Political Econom. Amer.. and consumers observe price without incurring any search cost. Consumer Res. 2004. Econometrica 69(6) 1403–1436. The effect of forced choice on choice. C. Marketing Res. Consumption ﬂexibility. given that in equilibrium we need to have p + t/ 2n > v. and the equilibrium has consumers with v ≥ v with v ≤ v incurring search costs k. D. Choice proliferation. Consider now the case where v is uniformly distributed. E. y ≡ −f p + + t 4n t − F p+ −F v 2n t 2n p− t 2n 2n t 2n f v 2p − v − f p+ t t 2n Note that for p = v − t/ 2n the condition (2) yields v = v. and market competition. 2006. 96(5) 1449–1476. which is positive as 2p > v. 1979. which implies y < 0.. and n. ﬁnite.. Variety for sale: Mass customization or mass confusion? J. Boatwright. one obtains d v/dn = z − a y − x /y and dp/dn = z − ax /y. 1990. R. Using (4) one can then obtain the condition for the optimal n as √ (5) 2t 2 − 6wt t + 5tw 2 − 4vnw 2 + 4w 4 = 0 √ where w is deﬁned as w ≡ nk Using (5) to write n as a function of w and taking the derivative of n with respect to w one obtains√ dn/dw < 0 as the condition n < t/ 4k translates into w < t/2 As dw 2 /dn = k + n dk/dn we have that dn/dk < 0 a higher search cost leads to a lower optimal number of products. A. as t/ 2n > kt/n. Hepburn. G.. However.. which implies 2p > v. Simonson. It would be interesting to investigate what happens when there is competition. Record 83(261) 191–203. AddisonWesley. 1338–1346. the ﬁrm will never charge a . 2007. R. 2006. This paper considers the monopoly case. Iyengar. p. J. J. Kahn. Marketing 65(3) 50–63. D. L. J. Proof of Proposition 3. C. Acknowledgments Comments from the department editor. Then we know that the optimal number of products is strictly below t/ 4k . 112(4) 848–886. including rights and permission policies. A contingency model for the selection of decision strategies. P. product conﬁguration... Dickson. But then a consumer with a valuation equal to v would not incur the search costs k. A model of price adjustment. Acad. 3(3) 439–449. is available at http://journals. Differentiating (2) with respect to v. L. one obtains x ≡ 2n/t f v 2p − v . St. v D/v = −2 nk/t. W. R. Gourville.” price below v. P. The price knowledge and search of supermarket shoppers. J. Working paper. Another interesting issue to investigate would be the implications on price discrimination and vertical differentiation of product evaluation costs. Marketing Sci. as noted in the text. A dualself model of impulse control. Steven Shugan. Note that v ≥ arg max p 1 − F p . pp. the ﬁrm gets strictly positive demand and proﬁt. Reducing assortment: An attributebased approach. the results extend to the case in which product quality is also a choice variable by the ﬁrms that consumers have to check. Diamond. 2001. B. 1998. two anonymous reviewers. F. Washington University. Adding x to the derivative of (2) with respect to p one obtains p+t/ 2n v References Beach. Econom. 1971. Gul. For search costs k (below k) converging to zero we know that the optimal number of products converges to inﬁnity. Marketing Sci. Kamenica. Then. MA. T. C. Econom.
Effects of quality and quantity of information on decision effectiveness. T.. choose. Working paper. Jacoby. E. Consumer Res. Ethics 52(2) 207–211. 14(2) 166–188. Price as a stimulus to think: The case of willful overpricing. University of California. Marketing Sci. Kamenica. Berkeley. A. Bus. . 1980.org/. Selling formats for search goods. Staelin. 2008. Experiment. RAND J. M. 26(1) 118–129. J. Jacoby. The cost of thinking. Marketing Res.informs. T. Rao. N. D. E. M. 7(2) 99–111. M. Information overload in a network of targeted communication. J.. Working paper. Econometrica 76(2) 263–305. M. 1994. Econom. L. Kuksov. Olson. Information load and decision quality: Some contested issues. B. H.. 2009. Bertini. L. Information overload with endogenous pricing. 2007. Simon. Less choice is better. J. Econom. University of Chicago. Consumer Res. G.1346 VillasBoas: Product Variety and Endogenous Pricing with Evaluation Costs Management Science 55(8). S. E. 1977.: Learn. J. Quart. J. 1977. ePub ahead of print July 23. Marketing Sci. J. M. When more alternatives lead to less choice. Marketing Sci. J. D. 4(1) Article 3.. Anticipating regret: Why fewer options may be better. Equilibrium price communication and unadvertised specials by competing supermarkets. 2004. Signalling price image using advertised prices. R. Greenberg. Psych.. Marketing Sci. W. 14(4) 569–573. choose: Emerging and prospective research on the deleterious effects of living in consumer hyperchoice. Choose. choose. Wathieu. 2006. Bus. Iyengar. Buyer search costs and endogenous product design. Agricultural Food Indust. J. 13(3) 298–308. Johnson.. Working paper. Psych. Sarver. Rev. 67(3) 345–370. Y. J. Lal. choose. G. J.. When choice is demotivating: Can one desire too much of a good thing? J. Broniarczyk. K. 2004. J. 1994. J. 2008. 2004. American Marketing Association. choose. Keller. 23(3) 304–316. B. VillasBoas. Rev. 1956. Marketing Sci. kidney market. The sound of silence: Observational learning in the U. Memory. 2008. N. 20(1) 61–81. sometimes. Berkeley. Kuksov. F. Econom. VillasBoas. 73–86. Chicago. J.. Contextual inference in markets: On the informational content of product lines. 79(6) 995–1006. Personality Soc.. S. Mick. 23(4) 490–499. J. Berkeley. Cognition 14(3) 534–552. © 2009 INFORMS INFORMS holds copyright to this article and distributed this copy as a courtesy to the author(s). University of California. Moving Ahead with Attitude Research. Econom. Wernerfelt. R. 2001. Psych. Decision making in informationrich environments: The role of information structure. Miller. M. 2007. pp. Van Zandt. Kamenica. Bettman. Lurie.0500v1. C. 1987... 1955. R. 2004. 2004. Shugan. S. A. 69(1) 99–118. choose. Berkeley.informs.S. Some implications of a more general form of regret theory. Chicago. Zhang. Marketing Sci. J. is available at http://journals. Payne. 98(5) 2127–2149. Matutes. 1338–1346.. Norwood. 2006. VillasBoas. J. R. Haidt. Amer. 1987. Lepper. Retail pricing and advertising strategies. Consumer Res. The magical number seven. Marketing Sci. 2000. Adaptive strategy selection in decision making. J. D. M. 1090. Wind. L. R. 1988. Organ. Syam. J. J. C. Contextual inference in markets: On the informational content of product lines. plus or minus two: Some limits of our capacity for processing information.journal. eds. 63(2) 81–97. 35(3) 542–560. 14(2) 200–213. 30(4) 473–486. R. http://mktsci. C. Communication strategies and product line design. Simester. J. Sugden. B. H.org/cgi/content/abstract/mksc. Theory 41(2) 270–287. 1995. including rights and permission policies. S. A behavioral model of rational choice. J. information processing perspective. Loomes. Additional information. G. Consumer response to price: An attitudinal.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.