Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Information and Monopolistic Competition

Author(s): Steve Salop


Source: The American Economic Review , May, 1976, Vol. 66, No. 2, Papers and
Proceedings of the Eighty-eighth Annual Meeting of the American Economic
Association (May, 1976), pp. 240-245
Published by: American Economic Association

Stable URL: https://www.jstor.org/stable/1817228

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1817228?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

is collaborating with JSTOR to digitize, preserve and extend access to The American Economic
Review

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms
INFORMATION AND MARKET STRUCTURE

Information and Monopolistic Competitio


BY STEVE SALOP*

Information gathering and transmission Thus, the average cost curve is U-shaped,
have been gaining a larger role in the attaining its minimum at (p*, q*).
economy over time. M. Porat (1975) has There are a large number of consumers
estimated that the production, processing L, who wish to consume this commodity.
and distribution of information goods and We begin by assuming that each consumer
services account for over a quarter of GNP. has an identical perfectly inelastic demand
George Stigler (1961), George Akerlof for one unit of the commodity at all prices
(1970), and the papers in Edmund Phelps's below pm. (Thus, pm can be interpreted as
volume (1970) showed the presence of im- the dollar valuation of the commodity's
perfect information gives firms market marginal utility. It is most convenient to
power at least in the short run and often in think of this commodity as a durable
the long run as well. Peter Diamond (1971) which is bought only once by each con-
showed that the existence of costly in- sumer.)
formation gathering led to an equilibrium A monopolistic producer surely charges
in which each small firm charged not the a price equal to pm. If there is free entry
competitive price but rather the monopoly into this market, new firms drive down the
price. Later papers have expanded Dia- price until it falls to the competitive price
mond's work for markets with different p*. No price dispersion is possible since
production and information gathering and each perfectly informed consumer will ap-
transmission technologies. T hese papers proach the lowest priced firm.
find that if information is costly, each
small firm obtains market power, and the II. Imperfect Information
equilibrium (if one exists) is characterized
In a Lancaster framework, the utility
by prices above competitive levels and
services derived from any commodity is a
sometimes price dispersion as well. The
complex function of the characteristics
relevant market structure with imperfect
that specify the commodity. The net util-
information is not perfect competition but
ity surplus equals the utility services
rather monopolistic competition.
derived less its price. Given a choice among
I. The Basic Competitive Framework competing commodities, the "best buys"
are those brands with highest net surplus.
There are a potentially infinite number
Initially, a consumer has imperfect ex-
of identical firms which can produce a
pectations regarding the net surplus of
homogeneous commodity from a produc-
each good. These imperfect priors derive
tion technology with fixed cost T and in-
from his general knowledge and past in-
creasing marginal cost function MC(q).
formation gathering and experience. It is
* Federal Reserve Board. These are the views of useful to think of this as a dynamic learn-
the author and should not be interpreted as the views
ing process. The prior can be thought of as
of the Board of Governors. I am grateful to my col-
laborator, Joseph Stiglitz, and L. J. Dillard, M. a probability distribution of the net sur-
Porat, and the editorial assistance of G. Edelen. plus of a commodity. The distribution may
240

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms
VOL. 66 NO. 2 INFORMATION AND MARKET STRUCTURE 241

be an unbiased estimate of the actual net expensive nondurables with unknown qual-
surplus (rational expectations) or may be ity that are consumed each period. Experi-
biased due to incorrect extrapolation of ence information is generally incomplete,
the past, unseen commodity specification since every commodity is not sampled,
changes, past sampling realizations, false and, in addition, the actual performance of
claims, or general ignorance. The forma- a brand may be a random variable. By
tion of the prior is made more difficult "search," we mean sampling before pur-
since the relevant questions may not even chase by studying consumer magazines,
be known until after some information is visiting or phoning stores, reading ad-
gathered. vertisements, etc. Price information may
The cost of information gathering will be generated by search. In principle, search
depend on the technology of information may yield complete information. For ex-
production and diffusion, the type, com- ample, one could envision a magazine like
plexity and number of commodities, and Consumer Reports that publishes all the
the consumer's priors and preferences. relevant information; a truthful guarantee
These will define the speed of the learning has much the same effect.
process. Information gathering improves There are important interactions (ex-
the quality of the posterior distribution. ternalities) among consumers in the pro-
The expected value of information can be cess. For example, if a consumer (having
viewed as the increase in expected net sur- had one economics course) thinks that
plus from the information. The effective search by other consumers keeps the mar-
cost of information includes its resource ket honest so that price reflects quality, he
and nonpecuniary costs and will depend on need gather no information himself but
the consumer's cost of time, preferences, will purchase the commodity with the
analytic ability, experience and under- greatest market share (John Conlisk and
standing of the market. For example, bar- Dennis Smallwood). Alternatively, if a
gain hunters and (presumably) economists consumer believes in the Lemon's Prin-
have low costs. ciple, he will drop out of the market alto-
Depending on its cost, complete or in- gether (Akerlof).
complete information may be gathered. By Before analyzing this market, let us ask
complete information, we mean informa- the question of why low cost consumers do
tion that enables the consumer to purchase not arbitrage the information differences
the best buy available. By incomplete, we away. Stated another way, since the mar-
mean that his expected net surplus will lie ginal cost of distributing the information
somewhere between the average and the is nearly zero, why won't competition
best buy. For incomplete information among information producers drive down
gathering, the consumer must choose the the price? First, consumers face the same
sample size, minimum acceptable net sur- information problem with respect to in-
plus level, or reservation price. formation sellers as they do with respect
Nelson (1970) contrasts two modes of to the commodities themselves. Each
information gathering, "experience" and magazine must be evaluated and its con-
"search." By "experience," we mean the tent processed and analyzed. Furthermore,
knowledge obtained from actually using since the dissemination of the information
the commodity. This information may be by each magazine exhibits decreasing
gathered from personal use or from the ex- average costs, the information industry
perience of acquaintances (shared informa- cannot be competitive in long-run equilib-
tion). It is more likely to be utilized for in- rium.

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms
242 AMERICAN ECONOMIC ASSOCIATION MAY 1976

III. Equilibrium

We now analyze the behavior and equi-


librium price structure of this commodity
market. Initially, the dynamics of the
learning process are ignored. An equilib- p =Pn-~-g-4-
rium in this market is n-prices p= (pi,
P2___ /
P2, . *. , pn) each charged by a proportion
/3= (p13 032, * , *3,^) of the N-firms in pro-
duction such that the following price and
entry competition conditions are satisfied: qn q2 q

FIGURE 1
(1) Price Competition (Maximum Profits).
Each firm chooses the price (and quan-
price be above the monopoly price. The
tity) which maximizes its profits given
exact specification of the equilibrium will
its knowledge regarding the behavior of
depend on the production technology, the
consumers and other firms (i.e., its de-
information gathering technology, and the
mand curve).
joint distribution of consumers, utility
(2) Entry Competition (Zero Profits). No
valuations and search costs.
new firm wishes to enter the market.
A simplified example can highlight the
The exact specification of a firm's de- properties of the equilibrium. Assume that
mand curve depends on the informational L-consumers all have identical valuations
and strategy assumptions we make about pm, known with certainty. Prior price ex-
firms. We will assume that a firm chooses pectations are rational but limited. Con-
a price given (i) tbe prices chosen by the sumers know the correct set of prices and
other (N-1) firms and (ii) the search rules proportions { p, d }, but a priori, they lack
of consumers. Thus, a firm is a "Nash" the knowledge of which firms charge which
competitor vis-a-vis other firms and a price. (This assumes away lemons and
"Stackleberg" competitor vis-a-vis con- two-armed bandits.) For a cost ci, con-
sumers. Condition (2) is the free entry as- sumer i can purchase complete information
sumption. We assume that entry occurs if from Consumer Reports in the form of the
profits are positive; hence, it is the zero complete price-location correspondence.
profit condition. There are only two groups of consumers,
In order to demonstrate the existence of a proportion a with cost cl> 0 and a pro-
the equilibrium, we follow the method of portion (1-a) with higher cost c2>cl. In
proposing a potential equilibrium set of spite of overwhelming empirical evidence
n-prices p= I Pi, P2) . .. p. 1, proportions to the contrary (A. Tversky), we assume
jIY= (a', * * *, |x outputs q= (ql, q2 ....consumers gather information in a rational,
q), and a number of firms-NS, and then maximizing
see manner.
if any of the existing N-firms wishes to If both groups have perfect information
change its price or if any potential entrants (c1= c2= 0), the competitive equilibrium
wish to enter. A potential equilibrium is an (p=p*) will obtain. If there are enough
actual equilibrium if both deviant be- perfectly informed consumers (cj =O, a
havior and entry are unprofitable. An high), the competitive equilibrium can still
equilibrium can be pictured as in Figure 1. obtain in spite of c2 > 0. This is an example
The minimum price (pi) will never be of the externality informed consumers give
below the competitive price, for that would to the uninformed. The weight of their
yield negative profits. Nor will the highest search keeps the market competitive.

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms
VOL. 66 NO. 2 INFORMATION AND MARKET STRUCTURE 243

If both groups have positive cost, the p'


competitive equilibrium cannot occur. A
pL
deviant firm could raise its price slightly P

above the competitive price and lose no


P*
customers. The only possible single-price
equilibrium is the monopoly price, with en-
time
try competition driving profits to zero at
FIGURE 2
this high price (Diamond). That equi-
librium will obtain only if information p* and the limit price. The exact dynamics
costs are large for both groups. At this will depend on the dynamic learning pro-
equilibrium, in order to induce consumers cess by consumers and firms. In general,
to pay the cost of discovering its location, prices may creep up slowly to above the
a deviant firm would have to follow the limit price which induces a price war down
unprofitable strategy of charging a price to p*, only to again begin the upward
below the competitive price. creep. (See Figure 2.) The frequency and
If lowering price to induce search is regularity of the cycle will depend on the
profitable, a two-price equilibrium (TPE) time adjustment speeds of price changes,
will obtain in which a proportion f of the entry, and the learning by consumers.
firms charge the competitive price (pi = p*) An interesting example of this process is
and the rest charge a higher price no greater that of dynamically captive markets. Sup-
than the monopoly price (pm>ph>p*). pose consumers have better information
Every type-1 consumer purchases com- regarding the brand they are now consum-
plete information and pays the lower price. ing than about other brands they con-
TI he higher price has the property that the sumed in the past or have never consumed.
type-2's do not find it worthwhile to pur- This asymmetry in information gives each
chase information. However, a proportion firm market power over its current risk-
d of the type-2's are lucky and randomly averse customers since there are effectively
select a low-priced firm. Entry competition "moving costs" of uncertainty from chang-
occurs until every firm makes zero profits. ing brands. If these moving costs are high
When only complete information can be enough, every firm can act like a complete
purchased, a TPE obtains even when there monopolist over its segment of the market.
are many types of consumers (Salop and Some deviant firm may find it profitable to
Stiglitz). In contrast, when incomplete in- subsidize these moving costs by charging
formation can be purchased, there will be a lower price. It gains customers who in
as many prices in equilibrium as there are turn become its captive as the information
types of consumers (Gerald Butters, Peter from their past experience decays. Hence,
von zur Muehlen). the deviant firm can creep its own price
If costs have little dispersion and are not upwards towards the monopolv price.
too high, there is no TPE either. However, Y. Shilony shows that a mixed strategy
a single "limit price" pL exists with the equilibrium in which each firm changes its
following property. A deviant firm could price every period results from such a
increase its profits by raising its price "captive market model," even when the
above pL . But if it does act in this short- commodities are nondurables bought every
run interest, it will destroy the equilibrium period. No firm's price ever settles down,
by inducing other firms to drastically cut so imperfections in information remain
price to induce search. Prices may begin forever. This analysis also has the interest-
to oscillate between the competitive price ing feature that the firms could form a

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms
244 AMERICAN ECONOMIC ASSOCIATION MAY 1976

able in the short run, and in the long run


Icycle p II C=1 C2 too if there are new suckers (ignorant con-
I limit I p sumers) born every minute.
[ Pi r)ipricci
V. Price Discrimination
pr=p* !/
ita-hlghow
So far we have focused on price disper-
TPE P"'P> __cyl
sion arising from search cost dispersion in
Pi or~~~i u which poorly informed consumers pay
price
higher prices. When consumers differ in
p =
their valuations as well, monopolistic com-
FIGURE 3 petition will generate price discrimination
against consumers with both high costs
C"noncooperative cartel" that needs to col- and high valuations. Low valuation-high
lude only once to set the random pricing cost consumers will not purchase. A mo-
rules. Once the cartel is begun, no single nopolist may also use dispersion in search
firm will ever find it profitable to cheat by costs to price discriminate against uniden-
deviating from the rule. tifiable consumers with inelastic demand
The various types of equilibria discussed curves (high pm's here). If he permits price
are summarized in Figure 3. dispersion (through unadvertised specials
or contrived heterogeneity), the high cost
IV. Advertising consumers will search less and hence pay
The firms themselves may engage in in- higher effective prices on average. If these
formation transmission through truthful customers have more inelastic demand,
advertising and guarantees. This does not then price dispersion acts as a price dis-
change the basic result. For example, sup- criminating tie-in of search (a "bad") with
pose expensive large ads and inexpensive the commodity he produces (Salop).
small ads are available, type-i consumers
VI. Competition vs. Welfare
are likely to notice both sizes, and type-2
consumers are likely to see only large ads. The existence of many similar brands
Then, firms that run large ads incur higher tends to increase effective search costs.
fixed advertising costs, charge a higher Suppose a consumer must purchase a
price and sell only to type-2's. On the brand in order to gauge its utility (an
other hand, suppose a firm sends ads out "experience" good). Optimal search would
randomly, and each consumer chooses the consist of sampling brands without replace-
firm with the cheapest advertised price ment until an acceptable one is found. In
from the ads received. In this model, price this case, entry competition has two coun-
dispersion results even if consumers are tervailing effects. The usual entry effect
identical (Butters). In both approaches, increases choice and decreases prices. On
the highly advertised brands have higher the other hand, in sampling without re-
costs and charge higher prices in equilib- placement, small population sizes are pre-
rium. Of course, this equilibrium result is ferred to large ones. For example, suppose
different from the short run practice of a there are 3 firms and a consumer evaluates
low priced deviant advertising its lower the net surplus of these 3 brands as { 5, 10,
price. False advertising, bait-and-switch, 15 }. If he samples without replacement 3
dishonored warranties and other forms of times, he will certainly attain the maxi-
fraud have also been assumed away in this mum utility level of 15. If the number of
analysis, though they are certainly profit- firms doubles to 6 and the distribution of

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms
VOL. 66 NO. 2 INFORMATION AND MARKET STRUCTURE 245

utility levels remains identical (i.e., {5, 5, M. Porat, "The Information Economy," Inst.
10, 10, 15, 15 }), then a sample of size 5 is for Communication Research, Stanford
1975.
required to obtain a level of 15 with cer-
M. Rothschild, "Models of Market Organi-
tainty. A sample of size 3 will give ex-
zation with Imperfect Information," J.
pected maximum utility of only 13.1. Thus,
Polit. Econ., 1973.
the tradeoff between search intensity and
, "A Two-Armed Bandit Theory of
utility level worsens, just as it would from Market Pricing," J. Econ. Theory, 1974.
an increase in effective search costs. In S. Salop, "The Noisy Monopolist: Imperfect
equilibrium, the increased cost tends to Information, Price Dispersion and Price
raise prices, ceteris paribus. The net effect Discrimination," Federal Reserve Board
of increased competition may be either to 1973.
raise or lower prices. Thus, there may be a and J. Stiglitz, "Information, Price
tradeoff between competition an(d static Dispersion and Monopolistic Competition,"
welfare. Federal Reserve Board 1975, Rev. Econ.
Studies, forthcoming.
Y. Shilony, "Mixed Pricing in Locational
REFERENCES
Oligopoly," Berkeley 1975, Rev. Econ.
G. Akerlof, "The Market for 'Lemons': Studies, forthcoming.
Qualitative Uncertainty and the Market D. Smallwood and J. Conlisk, "Product
Mechanism," Quart. J. Econ., 1970. Quality in Markets Where Consumers are
G. Butters, "Equilibrium Distributions of Imperfectly Informed and Naive," Univ. of
Sales and Advertising Prices," Princeton California, San Diego 1975.
1975, Rev. Econ. Studies, forthcoming. G. Stigler, "The Economics of Information,"
P. Diamond, "A Model of Price Adjustment," J. Polit. Econ., 1961.
J. Econ. Theory, 1971. A. Tversky, "Assessing Uncertainty," J. Amer.
P. Nelson, "Information and Consumer Be- Statist. Ass., 1974.
havior," J. Publ. Econ., 1970. P. von zur Muehlen, "Sequential Search and
E. Phelps, et al., Microeconomic Foundations Price Dispersion in Monopolistic Compe-
of Inflation and Employment Theory, New tition," Federal Reserve Board 1975, Rev.
York 1970. Econ. Studies, forthcoming.

This content downloaded from


202.43.95.117 on Mon, 05 Oct 2020 12:23:07 UTC
All use subject to https://about.jstor.org/terms

You might also like