Content uploaded by Robert Rothschild

Author content

All content in this area was uploaded by Robert Rothschild on Nov 25, 2014

Content may be subject to copyright.

34

The Theory of Monopolistic Competition:

E.H. Chamberlin's Influence on Industrial

Organisation Theory over Sixty Years

by R. Rothschild*

University of Lancaster, Lancaster, UK

Introduction

In 1933, Edward H. Chamberlin published the Theory of Monopolistic Competition

(1962).

The work, based upon a dissertation submitted for a PhD degree in Harvard

University in

1927

and awarded the David A. Wells prize for

1927-28,

has since become

a milestone in the development of economic thought. Its impact on industrial

organisation theory, general equilibrium and welfare economics, international trade

theory and, to a greater or lesser degree, all other branches of economic analysis, has

been pervasive and enduring. The ideas set out in the book have been developed,

expanded and refined in ways too numerous to be identified precisely, and the books

and articles which take Chamberlin's contribution as a starting point arguably exceed

in number those on any other single subject in the lexicon of economics[l].

Its status in the history of economic thought notwithstanding, the Theory of

Monopolistic Competition is in many ways a rather unsatisfactory work. It rests upon

assumptions which describe a world for which there is no empirical analogue: it seems,

in its original form at least, to be a solution in search of a problem. Indeed, from

a modern perspective, it would appear that the impact of the "Chamberlinian"

contribution is due less to the insights contained in the Theory than to the efforts

of those whose work has made up the research progamme which

was

set in train in 1927.

The purpose of this paper is to consider some of the links between Chamberlin's

ideas and the developments to which these ideas gave rise. In doing so, we shall hope

to place in perspective his early contribution to what is now known as "Industrial

Organisation" theory, and thereby to demonstrate the power and enduring influence

of his role in the evolution of economic thought.

Chamberlin's Contribution

The Theory of Monopolistic Competition deals with two types of market. The first,

to which Chamberlin gave relatively little attention in the book

itself,

involves a small

number of firms (oligopolists) who face a choice between myopic competition of the

kind first discussed by Cournot (1838), and joint-profit maximisation. Chamberlin's

* This paper was written in large part while the author was Visiting Fellow at the Australian Graduate

School of Management, University of New South Wales, Sydney.

Chamberlin and Industrial Organisation Theory 38

particular contribution was to show that the recognition of mutual interdependence

on the part of firms in the small numbers case is a necessary, if not sufficient[2]

condition for the attainment of a Pareto optimal outcome[3]. Although he chose to

relegate the discussion of the oligopoly case to a few pages, Chamberlin was in later

years to regard this market form as being of central importance in economic analysis

(see Chamberlin, 1957, 1961; Kuenne, 1967; and Skinner, 1983).

The second type of market which Chamberlin considered — and upon which we

shall focus in this discussion — is the "large group". It was in this context that he

sought to identify the key features of monopolistic competition. For the purposes of

his model Chamberlin regarded industries as being made up of "groups" of products,

each in turn being made up of

close,

but less than perfect substitutes. Groups could

themselves be distinguished from one another by the degree of substitutability of their

respective products, in this case taken to be smaller than that between varieties within

a given group. The classification of industries on this basis was not universally

accepted[4],

in part because of the marked pervasiveness and complexity of product

differentiation in practice. Even after six decades, the problems of definition and the

need to find a satisfactory basis for classification remain[5]. However, Bain (1967,

p.

153)

has observed that, despite its shortcomings, Chamberlin's conceptualisation has:

proved formally satisfactory, tractable and productive of meaningful hypotheses...by

assuming explicitly that the enterprise economy

is

made up of industries that

are

identified

and separated by the cross-elasticities of demand among products, and by then classifying

such industries according to their market structures

The elements of the large group model can be described with the aid of Figures 1

and 2. Let the industry initially comprise N firms, each producing a single variety

of the differentiated product. For the sake of simplicity, suppose that at any price

p the amount sold by any individual firm is 1/Nth of the total market demand. Let

costs be the same for all firms (in this case, the traditional U-shaped average total

cost curve is appropriate) and suppose that the prevailing price yields a surplus over

costs.

Suppose further that, initially, entry into the group is blocked. Under

Chamberlin's assumptions, each firm will perceive an opportunity to increase its profits

by lowering its price, provided that none of its competitors does the same. On the

basis of this

belief,

each firm will expect to increase its sales by moving down its

(relatively elastic)

ceteris

paribus demand curve

(dd);

but if all firms were to behave

in similar fashion, then each would find that its sales are given by the (more inelastic)

"share-of-the market" demand curve (DD). Chamberlin argues that the conjecture

of each firm to the effect that it can gain at the expense of its rivals will be rendered

invalid by the fact that all behave in this manner, each motivated by the belief that

its price reduction will have negligible (and equal) impact upon the sales of

all

others.

The result is that all firms find themselves on their respective DD curves. Yet none

is encouraged to revise its expectations: each reduces price still further, until ultimately

only normal profits are obtained.

When the possibility of free entry and exit is considered explicitly, the process,

starting at an arbitrary price above average total costs, is similar, except that now the

location of the DD curve is determined by the size of N. Super-normal profits attract

new entrants, and the DD curve shifts leftward at the same time as the individual

36 Journal of Economic Studies 14,1

firms undertake their myopic price reductions. The shift is halted when the prevailing

market price equals average total costs, but if at this point firms continue to perceive

opportunities for profitable price reductions, then these will take place. Eventually,

a point is reached at which all firms make losses, yet each conjectures that, provided

that none of its rivals behaves as it does, profitability will be restored through one

final price reduction. Since all firms entertain this naive expectation the result is that

losses are increased for

all.

The difficulties are resolved through the exit of

some

firms,

an action which has the effect of shifting the DD curve to the right. Eventually, a

Cournot equilibrium is attained where the dd curve is tangent to the average total

cost curve. At this point, no firm has an incentive to change its price, and neither

entry nor exit will take place. Here, a uniform

price,

equal to average costs, obtains for

all firms, but each produces an output smaller than that which would be produced if

Chamberlin and Industrial Organisation Theory 37

the dd curve were horizontal, as it is under conditions of perfect competition. This

phenomenon encouraged Chamberlin to the view that monopolistic competition would

give rise to a "waste" of

resources:

too many varieties will be produced, each on too

small a scale.

The

Theory

of

Monopolistic Competition

contains a substantial discussion of selling

costs.

The ideas put forward by Chamberlin are discussed in some detail in Abbott

(1955),

but much recent work in this area has departed from Chamberlin's original

treatment. For this reason, we shall omit from our review this aspect of the "large

group" case. However, it would be inappropriate to accord similar treatment to the

question of product variation as a competitive device, and much of the following

discussion will address this key issue.

For the purposes of this survey, we identify four aspects of the Chamberlin model

as it has been set out here:

38 Journal of Economic Studies 14,1

(3) the assumptions on the nature and consequences of "entry" (and "exit"); and

(4) the question of the "optimality" of product variety.

Symmetry and Myopia

Chamberlin's analysis rests heavily upon two interrelated assumptions. The first is

that any firm contemplating a price reduction expects to attract a very small proportion

of custom from each of

its

competitors; the second is that this proportion is the same

for all of those competitors. Consequently, every prospective price-cutter believes that

other firms will lose so small a proportion of their customers that none will respond

by cutting its own price. According to Chamberlin, this belief

is

correct, but because

each acts on this basis market price must fall. These two assumptions are themselves

closely linked with, and indeed provide the rationale for, Chamberlin's implicit

assumptions about the price cross-elasticities of demand for all pairs of products,

and their apparent insensitivity to changes in the number of varieties in the product

group.

There are two separate issues involved here. The first concerns Chamberlin's

assertions about symmetry and the negligible effect of firms' price reductions on the

sales made by their

rivals.

The second concerns the myopia to which these perceptions,

when shared by the agents themselves, give

rise.

These issues are of central importance,

because the assumption that firms are myopic in their pursuit of profits is justified

only if they consider that their actions have negligible impact upon each and every

one of their competitors. Any other conjecture raises the question of structural

interdependence, and the notion of comparative anonymity of the individual firm in

the large group is replaced by a more complex and less tractable set of relationships.

Dixit and Stiglitz (1977) have shown formally that the consumer-theoretic basis of

the Chamberlin model requires that demands be generated by an aggregate utility

function characterised by a constant elasticity of substitution between any pair of

varieties. As Nicols (1947) noted, this may be an unduly restrictive requirement. His

alternative is a formulation based on the assumption that customers have scales of

preferences, in terms of which they prefer one variety but are indifferent amongst all

others. In this case, the

ceteris

paribus demand curve for each firm can be shown

to have an obtuse kink at the going output, and Chamberlin's price cross-elasticity

assumptions, yielding as they do a continuous dd

curve,

turn out to imply a complex

and rather special combination of preferences on the part of customers.

An extension of the Nicols' model would be one in which the desire for variety

on the part of a customer is assumed to diminish continuously with the "distance"

of any given variety from his most preferred choice. The idea of distance as a proxy

for product differentiation has provided the basis for a substantial body of literature

on the subject of spatial competition. Since the appearance of the pioneering work

of Hotelling (1929)[6], theorists have recognised that a spatial representation is a

tractable and useful way of analysing aspects of the Chamberlin model. Chamberlin

himself recognised the potential of the spatial approach in his celebrated "Appendix

C" (1962), and also in his discussion of the general applicability of considerations

relevant to competition between small numbers

(1962,

pp. 103, 104). Because differences

in "location" may be considered a proxy for product differentiation in general,

Chamberlin and Industrial Organisation Theory 39

location patterns are analogous to configurations of "variety" in monopolistically

competitive markets. Consequently, if the distance between every pair of firms in the

spatial market were the same then, in Chamberlin's sense, all products in the group

would be equally substitutable for one another. The particular advantage of the spatial

representation is that it makes it possible to show the special circumstances under which

this will not be so, and the variety of configurations which may arise when different

assumptions are made. One of the crucial differences between the Chamberlin and

the Hotelling representations of a differentiated market has been identified by

Archibald and Rosenbluth (1975), and must be borne in mind. If

the

distances between

all pairs of varieties are not the same, then the impact of both price changes and entry

will be asymmetric and dependent upon the "proximity" of

given

varieties to the source

of the perturbation. We shall return to this point below, but it is worth noting here

that in this sense, the Chamberlin-Dixit-Stiglitz formulation of the underlying utility

function is less useful than that implied by Nicols.

In setting out some of the features of models of spatial competition which inform

and extend the Chamberlin analysis, we shall assume initially that the number of

varieties, N, is fixed. The most appropriate point of departure for those wishing to

link the work of Chamberlin with the literature of spatial competition is the "zero

variations" conjecture (ZCV). Under ZCV, each firm expects that its rival(s) will not

respond to any action on its part. In this sense, ZCV is the natural counterpart of

Chamberlin's assumption of myopia. Hotelling assumes a uniform distribution of

consumers' preferences over a line of finite length (a "one dimensional market")[7],

[8].

There are two firms, both identical in all respects but for their locations on the

line.

Letting both production costs and elasticity of consumers' demands be zero, he

assumes that each buyer purchases from the firm whose price plus transport cost is

the lowest. Hotelling allows the firms infinite mobility, so that each is able to adjust

its location until no further gain is possible given the choice made by its rival. The

equilibrium condition is thus identical to that of the Cournot model. On this basis,

he establishes some useful results for the two-seller case. The first is that the

introduction of

"space"

removes some of the discontinuities associated with traditional

non-spatial models of interdependence: if one firm undercuts its rival, then the latter

will not lose all of its custom. This perception conforms with Chamberlin's model.

The second result of interest is that, given identical prices for both firms, there can

be found a spatial equilibrium in which the two firms emerge adjacent to each other

in the middle of the market.

Recent work on the Hotelling model by D'Aspremont et

al.

(1979) has attempted

to show that under ZCV, too great a proximity between the two firms may prevent

a simultaneous equilibrium in both price and location from occuring. This result

contrasts with those obtained by Neven (1985, 1986) who demonstrates that a pure

strategy price equilibrium can be found for this case, and also for the case in which

the density of consumer demand is not uniform across the line. In the latter

formulation, the tendency of the two firms is however towards dispersion rather than

concentration at the midpoint. The two-firm model has also been considered by

Smithies

(1941).

In his framework, elasticity of demand is finite. His results show the

crucial role played by the elasticity assumption, and in particular the fact that in this

40 Journal of Economic Studies 14,1

case the tendency will be towards dispersion. In this sense, his result appears to offer

some support for Chamberlin's implicit assumption that, in differentiated markets,

varieties will be symmetrically dispersed[9].

Unfortunately, however, the introduction of larger numbers of varieties makes

matters very much more complex. Chamberlin (1962) has shown that there is no stable

(pure strategy) equilibrium in locations (given identical prices) when N =

3.

A similar

result has been proved by Shaked (1975) and Graitson

(1979),

although Shaked (1982)

has shown that, for this

case,

there does exist an equilibrium in mixed strategies. Lerner

and Singer (1937) and Eaton and Lipsey (1975) have shown that a purely locational

equilibrium can, however, be found for any

A7

greater than 3. The results confirm a

tendency towards dispersion, but in general the distance between adjacent firms is

not uniform. Other formulations of the N-firm problems under ZCV include

Carruthers (1981), who models a case in which each firm assumes that its rivals'

locations are fixed, but that they will adjust their prices to his choice of price and

location. This has the effect of bringing the firms located towards the end of the line

somewhat closer to the interior than is the case in the models of Lerner and Singer

and Eaton and Lipsey[10].

The analyses presented above are based upon a somewhat restrictive assumption.

Novshek (1980), for example, has observed that in spatial markets where marginal

costs are constant, ZCV may be an inappropriate concept: the fact that firms whose

markets overlap must always affect each other means that no equilibrium of interest

can occur. According to Novshek it is, not surprisingly, only monopolies (whose

markets are, by their nature, distinct from others) for whom ZCV is a legitimate

operational assumption, and for these firms it

is

largely irrelevant in any

case.

A similar

conclusion can be found in Kohlberg and Novshek

(1982),

who show that the existence

of equilibrium depends in a crucial way upon the length of the market relative to the

number of firms. Quite apart, however, from the restrictions necessary to secure

equilibria of interest, there is the objection raised by Kamien and Schwartz (1983),

who point out that the logic of ZCV is itself suspect: if each firm conjectures that

rivals will not respond to its actions, how does the firm justify its own response to

theirs? A wide variety of possibilities arises once the ZCV assumption is relaxed.

Gannon (1972) has argued that equilibrium in the two-firm case may be anywhere

in the market, depending upon the conjectures of the firms. His general principle is

that firms will emerge closer to each other if each believes that the other will respond

"weakly" (ie. to a smaller extent) to any change in its own location, and vice-versa

for "strong" expected reactions. D'Aspremont et

al.

(1979) and Graitson (1980) have

proposed a modified concept, the "maximim" conjecture, for which an equilibrium

in both prices and locations (each firm at the midpoint of its respective market) may

be found. Rothschild (1976, 1979) has investigated the application of the maximin

concept to the case where N

is

fixed and greater than 2, and demonstrates a general

tendency towards dispersion[ll].

Perhaps the most important, yet least developed, of the analytical issues arising

from an explicitly spatial representation of the work of Chamberlin is the problem

of "chain-linked" markets. A market may be said to be chain-linked if a cut in price

by one firm affects more strongly its proximate rivals, leaving relatively unaffected

those further away. Here, proximity encompasses both physical proximity (as in the

Chamberlin and Industrial Organisation Theory 41

case of gasoline filling stations) and similarity in characteristics (as in the case of

varieties of

cider).

Either

way,

the argument that proximate competitors will be more

affected by a price reduction, and hence will be more likely than others to respond,

is an intuitively and empirically appealing one. Kaldor (1935) was the first to raise

the objection that markets are typically chain-linked rather than symmetrical in

Chamberlin's sense. Chamberlin's symmetry assumption makes it unnecessary for him

to consider this important question, but he

shows

himself

to

be aware of

it

(Chamberlin,

1962,

pp. 103, 104). A brief discussion of chain-linking can be found in Copeland

(1940) and Henderson (1954), but the idea

is

not developed. Lancaster (1966) has dealt

with the concept in the course of setting out his "characteristics" approach to the

theory of demand. This framework, which Lancaster (1966), Baumol (1967), Salop

(1979) and Archibald, et

al.

(1986) have shown to be similar to the one-dimensional

spatial models, rests upon the assumption that consumers typically demand

characteristics of products rather than products themselves. Competition between firms

is thus seen to be competition between "bundles" of

characteristics.

The analogy with

spatial representations of horizontally differentiated products is readily apparent, and

writers in this area have emphasised the localised (non-symmetrical) nature of

competition in markets defined in this way,

Archibald and Rosenbluth (1975) have integrated Lancaster's approach to demand

with Chamberlin's large group model, and show that the latter's ceteris

paribus

demand

curve (dd) is an absurd construct if the products in question have fewer than four

distinguishing characteristics. The principal reason is that, in Lancaster's formulation,

such a situation would give rise to a series of chain-linked markets. However, if the

number of characteristics identifying each product exceeds four, then although the

likelihood of discontinuities in competitors' reactions is increased, the possibility that

these might be small or mutually offsetting is sufficient to provide a rationale for

Chamberlin's implicit assumptions. In similar vein, Capozza and van Order (1982)

suggest that, even when competitors are few, price reductions may not be followed.

In this sense, the effect of chain-linking may be less serious than might be expected.

The problem of chain-linking nevertheless remains one of quite crucial importance

in industrial organisation theory. As Friedman (1977, 1983) has argued, the concept

provides a sensible view of monopolistically competitive markets, and one which is

in many ways superior to that commonly found in the literature. However, although

the question is addressed in Friedman (1983), he offers few results apart from a

formalisation of some of the relationships. Perhaps the most detailed analysis of the

phenomenon to date can be found in Rothschild

(1982,

1986),

who shows, on the basis

of

a

particular set of

expectations,

how an equilibrium in prices may emerge in a chain-

market in which all adjacent firms are equidistant.

The natural extension of the analysis of chain-markets is to the question of the

effect of changes in N. In Chamberlin's formulation, the cross price-elasticity of

demand for each existing variety remains unchanged in the face of entry and exit.

The spatial analogy, and in particular the phenomenon of chain-linking suggests that

changes in the number of firms must be expected to change the relative proximities

of the available varieties. If this occurs, then individual price cross-elasticities must

also

change.

The problem remains an important and potentially fruitful one for further

research.

42 Journal of Economic Studies 14,1

Uniformity and Excess Capacity

Chamberlin's assumptions of uniformity in demand and costs have also received a

great deal of critical attention. Stigler (1950), for example, noted that product

differentiation and uniformity of cost and demand curves across all firms are mutually

exclusive concepts: differentiation, by its very nature, serves to make firms dissimilar,

and dissimilarities manifest themselves most commonly in non-uniform costs. The

point was not lost on Chamberlin, who made a number of attempts[12] to modify

the assumption. A recent contribution may have provided a basis for countering

Stigler's objection. Sher and Pinola (1981) have argued that a sufficient condition for

the existence of both of Chamberlin's demand curves (dd and DD) is that the prices

of products stand always in the same proportional relation to one another. Extending

this argument to cases in which the respective cost curves of the producers stand in

the same

proportional

relation to one another as do their

prices,

would make it possible

to identify an equilibrium for the group at some configuration of these prices.

This modification of the Chamberlin uniformity assumption, valuable though it

is,

does not take account of the effect of entry and exit. Irrespective of whether costs

and demands are identical or merely in

fixed

proportions to one another, the important

questions in this context are firstly whether group equilibrium will be characterised

by excess capacity in Chamberlin's sense and secondly whether the process will lead

to the zero-profit outcome predicted for the large group case.

We

turn first to the question of

excess

capacity. In a series of

papers,

Demsetz (1959,

1964,

1968, 1972) addressed Chamberlin's assertion that, in equilibrium, firms would

produce a smaller output than that necessary to minimise average total costs. The

essence of

Demsetz'

argument is that when a profit-maximising firm changes its output,

it will also attempt to shift its demand curve through increased selling expenditure.

When this occurs, the demand curve ceases to be fixed in the sense assumed by

Chamberlin. As an alternative to Chamberlin's demand curve, Demsetz introduces

the concept of a mutatis mutandis average revenue curve which depicts the optimum

price for each quantity sold. On the assumption that returns to selling expenditure

initially increase and subsequently decrease, the Demsetz curve acquires an inverted

U-shape. On the further assumption that firms will always wish (or be forced by

competition) to choose the most efficient available process, the tangency of

the

mutatis

mutandis curve with the average total cost curve ensures that the locus of market

equilibria will be horizontal, thus permitting the attainment of

a

zero-profit equilibrium

at the level of minimum average cots.

Although the Demsetz contribution has been extensively criticised (see for example

Archibald, 1961; Barzel, 1970, Perkins, 1972; Schmalensee, 1972; Ohta, 1977 and

Murphy, 1978) it has, together with the work of Dewey (1958) and Fama and Laffer

(1972),

formed the foundation for the development of a new and potentially exciting

research area known as the Theory of Contestable Markets. Dewey argued that

Chamberlin's tangency position would be inherently unstable. If competition is

imperfect, then rationalisation of the industry must be expected to lead ultimately

to production at minimum average costs. A necessary condition here, as Dewey points

out, is that entry into the market must be an instantaneous possibility. The leading

contributors to the contestability literature (Panzar and Willig, 1977; Baumol et al.,

1982 and Baumol, 1982, characterise a contestable market as one which permits

Chamberlin and Industrial Organisation Theory 43

absolutely free (and instantaneous) entry and costless exit. The entrant suffers no

disadvantages relative to incumbents and can, upon leaving the market, recover all

costs incurred on entry, subject, of course, to adjustment for any normal user costs

and depreciation. Since this condition serves to eliminate all risks involved in entry,

markets defined in this way are vulnerable, irrespective of the time period under

consideration. Because of

the

absence of barriers to entry and

exit,

contestable markets

are characterised by zero economic profits, no matter how many firms operate. In

such markets there exists no inefficiency, however defined, for the reason that

inefficiency constitutes an invitation to entry and will thus be eliminated. Moreover

price can never be less than marginal costs, since potential entrants could, in those

circumstances, obtain higher profits than incumbents by shading price and eliminating

unprofitable units. It follows that, in contestable markets, price must always equal

marginal costs, so that the necessary conditions for Pareto optimality are satisfied.

The results reported here introduce explicitly the question of entry, and the issue

of whether it will give rise to equilibrium. This is itself an important subject and,

despite the vigour with which the contestability argument has been pursued, by no

means a settled one. Even if the symmetry and uniformity assumptions were valid

in general, the likelihood of a zero-profit group equilibrium under free entry and any

imaginable conjectures on the part of firms would remain a separate and controversial

issue Kaldor (1935) argued that zero-profit equilibrium might not occur in the presence

of indivisibilities and economies of

scale.

In this case, entry will not lead ultimately

to a Chamberlinian tangency, but rather, will cease prematurely because of high costs.

Using the spatial analogy, the "interpolation" of a producer "between" any two others

might actually cause profits to become negative, thus deterring entry and yielding

monopoly advantage to pre-existing firms. This argument has found broad acceptance,

although Capozza and van Order (1982) have observed that it is weakened somewhat

if capital is mobile and the market is large relative to the indivisible unit of capital.

However their results also show that if capital is not mobile, then Kaldor's objection

cannot be so easily dismissed. A similar conclusion can be found in Eaton and Lipsey

(1976,

1978). Peles (1974) also concludes that pure profits may exist in equilibrium,

and argues that the presence of untransferable fixed assets allows firms to enjoy

monopoly power. But he shows that the quantity produced by these firms may

correspond to that associated with minimum average costs, so that in this sense

production is efficient.

Similar ideas have been put forward by Tullock

(1965),

Telser

(1969)

and Salop (1979).

With the aid of a spatial analogy, Tullock shows that a Chamberlinian zero-profit

equilibrium is indeed possible, and that it will also be Pareto optimal if consumers'

tastes are taken explicitly into account. In the same vein, Telser distinguishes between

markets of finite and infinite

length.

In the former, entry will lead to a Chamberlinian

equilibrium. In this sense, the restriction imposed by the finite nature of the market

gives rise to inefficiency. However, where the market is infinite, efficiency can in

principle be achieved. Salop investigates the problem using the framework of a

Chamberlinian "group" and an "outside" market. In his formulation, consumers buy

either one unit or none of the differentiated product, and spend the remainder of

their income on the homogeneous outside

good.

The differentiated product

is

produced

44 Journal of Economic Studies 14,1

subject to decreasing costs; the other is produced under competitive conditions. For

this case there can be shown to exist a variety of zero-profit equilibria, two of which

are of particular interest. The first is a confirmation of the existence of the traditional

Chamberlinian equilibrium; the second is an equilibrium in which profits are zero

but the demand curve is kinked because of the conjectures of the firms involved. The

result is similar to that of Sweezy (1939), although the conjectures of the agents are

somewhat different. While Salop does establish conditions under which profits in

equilibrium will be

zero,

he also provides an insight into the circumstances under which

this outcome will not necessarily occur. In particular, he notes that the existence of

indivisible

fixed

costs may require that the number of varieties in the market be integer

valued, thus preventing a zero-profit equilibrium from being attained. In this sense,

his results may be seen as being in the tradition of Kaldor[13].

Some of the problems of indivisibility, monopoly and efficiency which arise in the

context of a partial equilibrium formulation of the Chamberlin model disappear when

the framework is that of a general equilibrium. Hart (1978, 1980, 1982) has argued

that a monopolistically competitive equilibrium can exist given correct conjectures

on the part of firms about the demand for all potential differentiated goods, and that

this equilibrium will be approximately Pareto optimal if the economy is sufficiently

"large".

This result

does,

however, turn crucially upon certain restrictions on consumers'

preferences. Novshek (1980) offers similar results to Hart on the basis of a modified

ZCV assumption. Gabszewicz and Thisse (1980) also address the relationship between

the size of the market and the nature of the resulting equilibrium. They show that

increases in the number of differentiated products ultimately bring about a perfectly

competitive outcome, provided that the firms choose prices non-cooperatively. But

there exists an upper bound to the number of firms which can be active in such a

market.

It will be clear from the foregoing discussion that no unambiguous conclusions

can be drawn as to the existence or otherwise of a zero-profit equilibrium under

monopolistically competitive conditions. The question of whether or not such an

equilibrium would be characterised by the existence of excess capacity merely serves

to complicate the

issue.

The mobility of capital will doubtless

be

a factor of importance

in both respects. But even if

capital is

perfectly

mobile,

the characteristics of equilibrium

prices will depend upon the conjectures of firms, the preferences of consumers and

the size of the economy.

Variety, Market Pre-emption and Welfare

The issues raised in this section relate to the optimal provision of variety from the

perspective of both firms and consumers. As already noted, Chamberlin gives explicit

attention to the question of

entry.

However, in his model firms already in the market

do not anticipate new competition and so make no provision for it when deciding

upon the location of the products in the spectrum of

varieties.

In practice, of course,

existing firms must be expected to devote considerable resources to the matter of

selecting varieties and, in particular, to pre-empting the choices of potential

competitors. The need to do so arises in large part from the commitment which a

particular choice of variety imposes upon firms. In Chamberlin's model the entry

Chamberlin and Industrial Organisation Theory 48

of new firms cannot be prevented and, once it occurs, the market shares of all

incumbents are reduced by equal amounts. Using the analogy of the spatial market,

the implicit assumption in this case

is

that firms are "infinitely

mobile"

in the Hotelling

sense, and can adjust to the new entrant by establishing an accommodating set of

equidistant locations post

entry.

In practice, of course, such possibilities are severely

restricted by the existence of relocation costs[14].

Once the possibility of entry is introduced explicitly, firms face the need to choose

product varieties subject to expectations of

the

behaviour of both existing and potential

rivals.

The anticipation of the choices to be made by subsequent entrants is a necessary

condition for securing optimal provision of variety ex post from the firms' point of

view. The analysis of the problem has proceeded along two distinct lines. The first

assumes that N is determined exogenously and that firms attempt only to ensure their

maximum attainable advantage given the certainty that the industry will contain that

number. The second assumes that N is determined endogenously and that firms deter

entry by means of appropriate choices of location on the variety spectrum.

Perhaps the earliest discussion of the first version of the problem is to be found

in Rothschild (1976, 1979), who considers a case in which N is exogenously determined

but firms enter sequentially. All firms are assumed to sell at the same price, so that

only location is a choice variable. The market is taken to be the circumference of a

circle, with each consumer buying from the nearest seller. Firms choose their locations

on the assumption that succeeding entrants (whose number is known with certainty)

will choose the most unfavourable location from their point of view; each assumes

further that all rivals entertain precisely the same maximin conjecture. The pattern

of locations which results depends upon the elasticity of demand for each variety:

if elasticity is zero, then the first N-l entrants will locate equidistant, while the Nth

will choose a location arbitrarily; if elasticity is finite, then all firms emerge more

or less at the midpoints of their respective markets, but the early entrants obtain larger

markets than those who enter later. In both models there is a tendency towards a

dispersion of varieties over the variety spectrum.

The second of the two lines of analysis has been dealt with by Hay (1976), Prescott

and Visscher

(1977)

and Lane

(1980).

All analyse the strategic aspects of entry deterrence

under sequential entry. In Hay's model, new entry is prevented by a process of brand

proliferation, in which firms leave insufficiently large gaps in the spectrum of product

varieties to support newcomers. In Chamberlinian terms, the profits obtainable by

new firms are less than those at the tangency of average revenue and costs. When

entry is deterred in this

way,

the general tendency

is

for incumbents to disperse rather

than cluster, but in this they are constrained by the need to ensure that the interstitial

market segments are kept sufficiently small. The important insight obtained by Hay

is that it is minimum market size rather than price which constitutes the effective

deterrent to new competition. In his formulation, once entry has been deterred, the

prices of all firms adjust so as to maximise the profits of

each.

He shows that similar

conditions apply, and similar market configurations emerge, when the market is

growing. Moreover, the analysis can be shown to proceed as before even when demand

is not uniformly spaced, except that under these circumstances the spacing of firms

in equilibrium will be irregular.

46 Journal of Economic Studies 14,1

Prescott and Visscher consider a number of different formulations of the sequential

entry problem and show that, for each, the tendency is towards a dispersion of varieties.

When N is

fixed,

a Hotelling-type approach yields a result similar to that of Rothschild

(1976);

when waiting time is introduced explicitly, a Cournot equilibrium in both price

and location is obtained; when there is waiting time and prices can be chosen (and

costlessly varied) after the choices of others have been observed, then there also exists

an equilibrium; finally, when production capacities are chosen sequentially under

conditions of restricted entry and firms are permitted to choose plant in any number

of locations, complete monopoly results. The idea of plant proliferation as a tactical

device is also explored by Peles (1974). Results of the latter type are of some interest

when the social optimality of product variety is being considered. We return to this

topic below.

Lane offers an analysis which draws on elements in the work of both Hay and

Prescott and Visscher. On the assumptions that firstly each consumer has a unique

set of preferences and secondly both prices and locations are endogenously determined,

he sets out a model in which firms enter sequentially and cluster in the centre of the

market. A firm's optimal location will be found at the point where there exists a balance

between the tendency to cluster in this

way,

and the pressure to disperse which results

from competition in

price.

The patterns of location which characterise Lane's sequential

entry equilibrium yield to early entrants higher profits than those obtained by others.

The level of fixed costs for the firm will be a factor in the decision to deter new entry:

if these costs are high, then entry deterrence is possible; if they are

low,

then the profit

maximising strategy for the firm may require that new entrants be accommodated.

A similar result has been obtained by Dixit (1979) in a somewhat different context.

The natural extension of the arguments contained in some of the work outlined

here

is

the use of brand proliferation as an element of strategic behaviour. A discussion

of the issue can be found in Archibald and Rosenbluth

(1975)

and Schmalensee (1978),

who uses the principle in his analysis of entry deterrence in the "ready-to-eat" breakfast

cereal industry. Recent work of a similar kind is to be found in Lyons (1986).

In turning to the question of the optimality of product variety from the point of

view of the consumer, we address the welfare implications of monopolistic competition.

Some early writers, such as Kahn (1935) concerned themselves exclusively with the

relationship between price and marginal costs which characterises the equilibrium of

the large group. Kahn argued that, where price exceeds marginal cost, Pareto optimality

can still be achieved if the ratio of these variables is uniform throughout the economy.

But, as Bishop (1967) has observed, this approach to the question of optimality neglects

the important relationship between variety and welfare, which in his view remains

indeterminate as long as Chamberlin's uniformity assumption is retained, even if other,

more traditional optimality conditions are met. The problem, although partly eased

if the uniformity assumption is abandoned, is never entirely eliminated. In similar

vein, Kaldor (1935) has argued that it is impossible to draw any general conclusions

about the implications of excess capacity for welfare.

Chamberlin's own views on the welfare implications of large group equilibrium are

to be found in a number of

sources.

His exposition in Chamberlin (1950) is perhaps

the best known of

these.

There he observes that the consequences of monopolistically

competitive elements had either been "ignored or seriously misunderstood", and that

Chamberlin and Industrial Organisation Theory 47

the view of perfect competition as an ideal (as, for example, in Kahn, 1935) reflected

an underestimate of the pervasiveness of monopoly in the real world and a consequent

failure to incorporate it into a workable definition of welfare. In essence, whenever

diversity is demanded, marginal cost pricing

ceases

to be a basis for welfare judgements,

and comparisons of the state of affairs in large group equilibrium

—

which he terms

"a sort of an ideal" — with the prescriptions of the perfectly competitive model are

therefore meaningless. Chamberlin is equally clear about the difficulties involved in

establishing the optimality of product variety. In his

view,

no unambiguous statements

are possible a priori.

Whilst no recent contributions to the subject have developed the issues raised by

Kahn (1935), a number of writers have attempted to show that, in monopolistically

competitive markets, variety will be over-, under-, or optimally-provided[15]. Lewis

(1945),

for example, argues that monopolistic competition may lead to either excess

capacity or optimal provision, depending upon the circumstances surrounding the

particular class of

trade.

Spence (1976a) suggests that product selection failures ("too

many, too few, or the wrong products") may be a significant part of the overall costs

of market imperfection, but concludes that these costs are not currently measurable.

In Spence (1976b) he shows that, given monopolistically competitive pricing, high own

price and price cross-elasticities are likely to lead to the production of too many

products, while the converse is probable when both of these elasticities are low. A

similar result can be found in Dixit and Stiglitz (1977), who distinguish between the

"market" solution and the socially optimal outcome. The basis of their analysis is

the conflict between the objectives of producers (profit maximisation) and those of

the consumer (consumer's surplus), and they demonstrate that under certain conditions

on consumers' utility functions free entry will lead to the provision of a less than

optimal number of varieties[16].

The relationship between economies of

scale

and the extent of product differentiation

has been investigated by Meade

(1974).

He recognises a general need for variety, which

he regards as being greater the smaller the degree of substitutability among products.

But he argues also that the benefits of variety must be set against the opportunity

costs involved in its provision, measured here in terms of

the

economies of

scale

which

derive from concentration on the production of a narrower range of products. He

observes that, as a general

principle,

if costs of production

are

high

while

substitutability

is low, then firms will concentrate upon a smaller number of varieties. In such

circumstances, cost considerations will dominate the need for substitutability even

though the socially optimal solution demands variety. Koenker and Perry (1981)

consider circumstances under which scale economies will give rise to excessive variety.

The problem has been approached from an explicitly spatial perspective by Stern

(1972),

who shows that monopolistic competition may lead to market areas of above

optimal size (too little variety) as well as to market areas of below optimal size (too

much variety) according to circumstances. In similar vein, Heal (1980) has observed

that small markets will be under-served in terms of variety, while the reverse is true

for larger markets. In this sense, the extent to which variety is provided is determined

by the size of the market.

Another approach to the question of optimal variety stresses the importance of

market structure in general. Lovell (1970) argues that both price and variety may be

48 Journal of Economic Studies 14,1

inappropriate in some

sense.

In particular, to the extent that the variety being offered

reflects the dominant market form and also the conjectures of the firms involved,

socially optimal provision

is

unlikely except in unique circumstances. Swan

(1970)

shows

that a monopolist operating under conditions of constant returns to scale will offer

optimal variety. On the same subject, but for a more general case, Lancaster (1975)

has shown that, under imperfect competition and increasing returns to scale,

differentiation will not in general be socially optimal, but the precise direction of bias

depends upon the particular market structure. In his view, if returns to scale are

increasing, then monopolistic competition gives rise to excessive differentiation, and

monopoly to too little. However, if returns to scale are constant, then market

imperfections are not in themselves a cause of sub-optimal provision of

variety.

The

inference for both writers is thus that market conditions do not matter when constant

returns prevail. The point is disputed by White (1977), who shows that a monopolist

working under such conditions will offer optimal variety only if allowed to discriminate

on the basis of

price.

Otherwise, some of the products desired by consumers may not

be offered. Using a similar argument, Gabszewicz (1983) has shown that, even if

overhead costs are excluded, a monopoly may produce less variety than is socially

optimal.

The review of the literature in this section suggests that no firm conclusions should

be drawn as to the optimality or otherwise of product provision in differentiated

markets, from either the producers' or the consumers' points of view. Whether or

not variety will be over- or under-supplied depends upon the technical conditions in

production, the structure of the market, the conjectures and strategic behaviour of

firms,

and the utility functions of the consumers.

Concluding Comments

The nature and significance of the Chamberlinian contribution are best judged in

terms of the literature which his original work has inspired. On this criterion, as our

brief review will have demonstrated, Chamberlin's impact upon the evolution of

economic analysis must be beyond

dispute.

Although his view of

the

world, as expressed

in the Theory of Monopolistic Competition, has been shown by subsequent writers

to have been overly simple and, in places, crucially flawed, the insights upon which

it was based have themselves formed the foundation for much of what is today called

Industrial Organisation theory. Viewed in this light, Chamberlin's contribution must

be regarded as a cornerstone of economic analysis and a guarantee that his name will

enjoy a permanent place on the roll of the great economic theorists.

Notes

1.

Although Joan Robinson's celebrated book, The

Economics

of

Imperfect

Competition (1933),

appeared at the same time

as

the work of Chamberlin, it

is

undoubtedly the latter which

has

received

more attention. In this review we shall address ourselves exclusively to his contribution. For a

discussion of

the

intellectual background to Chamberlin's work see Chamberlin

(1961)

and papers

in Kuenne (1967).

2.

See Fellner (1967) for a discussion of this point.

3.

See Friedman (1977) for an analysis of the conditions under which noncooperative behaviour may

yield outcomes which are Pareto optimal, and Formby and Smith (1979) for an exposition and

critique of Chamberlin's cooperative outcome in the two-firm case.

Chamberlin and Industrial Organisation Theory 49

4.

See Bain (1967).

5.

See Bishop

(1952),

Heiser

(1955),

Bishop

(1955),

Fellner

(1953),

Chamberlin

(1953)

and Bishop (1953).

6. Similar analysis can be found in Launhardt (1885) and Zeuthen (1929).

7.

This particular form has been widely used as an analogue for monopolistic competition, even

though in such a market all firms can never be equidistant from all others. The point, however,

is that the absence of symmetry of this kind from the one-dimensional market must imply its

absence from an N-dimensional one.

8. The basic Hotelling model has attracted some criticisms which

we

shall not develop

here.

Hartwick

and Hartwick (1971) have shown that the equilibrium price in such a market may not be unique,

but will depend upon the initial price and location choice of the first firm to act. Devletoglou

(1965) bases his characterisation on a slightly different set of

assumptions.

He identifies a "minimum

sensible" constraint of indifference on the part of

buyers.

Given the existence of such a constraint,

it can be shown that firms will be repelled from the centre.

9. Models similar to that of Smithies have been investigated by Lerner and Singer (1937), Webber

(1972),

Eaton and Lipsey

(1975),

Graitson

(1979)

and Economides

(1982),

who assume rectangular

demand functions with the particular property that consumers buy one unit of the product up

to some

price,

and nothing above

it.

In this

case,

depending upon the size and shape of

the

market

area, different equilibria can be found, their number and character depending, in turn, upon the

number of firms in the market. Vickrey (1963) has considered the Hotelling/Smithies problem

in the context of a circular, rather than a linear market, and confirms the tendency to dispersion

when elasticity of demand is finite.

A number of writers have argued that the tendency will be toward agglomeration, rather than

dispersion. Lewis (1945) has suggested that larger retail firms will tend to cluster together. Stahl

and Varaiya (1978) attribute such behaviour to imperfections in information on the side of firms,

while Stuart (1979) argues that clustering is more likely to be due to uncertainty on the part of

consumers. Although we shall not develop these issues here, their importance for the theory of

location, especially retail location, should be noted.

10.

Ali and Greenbaum (1977) consider a similar case involving sequential entry into the market for

banking services.

11.

Teitz (1968)

shows that, if

firms

are permitted to

own

multiple

plants,

then an equilibrium

in

locations

will only be likely to occur if maximin location strategies are adopted. Variations on this game-

theoretic approach include Stevens (1961), who models the Hotelling/Smithies framework as a

zero-sum game, and shows that if the location points are discrete and choices simultaneous, then

their results hold. Gal-or (1982) considers the possibility of mixed strategies, showing that when

firms choose a distribution of

prices

(rather than a single

price),

equilibria may exist where otherwise

none would be possible. Kohlberg (1983) has shown that if the Hotelling problem is modeled as

an N-person game in which strategies are location choices and the payoffs are market shares, then

relocation gives rise to jump discontinuities. But if consumers choose outlets on the basis of travel

time plus waiting time (the latter being a function of the market shares of firms) then market

shares become continuous functions of

the

firms' locations. Unfortunately, no equilibria exist for

the case where N > 2.

12.

See, for example, Chamberlin (1962, p. 82ff).

13.

The work of

Losch (1954) is in

many

ways

a parallel

to

that of Chamberlin.

Using

a spatial framework

(in this case, an areal market), Losch shows that a system of regular hexagonal markets, one for

each firm, yields a unique zero-profit equilibrium under conditions of free entry. In contrast to

these findings, and on the basis of rather different assumptions on costs, Mills and Lav (1964)

have shown that there may be a wide variety of market shapes consistent with free entry, so that

profits in equilibrium may not be

zero.

Extending their analysis to include the question of efficient

allocation, and comparing their

findings

with those of Chamberlin, they conclude that equilibrium

may be characterised by a misallocation of resources especially where de-centralised processes are

involved. (A discussion of this point can be found in Samuelson, 1967; and Grace, 1970). Eaton

(1976) and Eaton and Lipsey (1978) provide a rationalisation and confirmation of the Mills and

Lav "positive profit" argument in the context of a one-dimensional market in which firms entertain

ZCV with respect to rivals' locations and the expectations that their respective mill prices will

not be undercut. Niedercorn (1981) offers a similar result.

30 Journal of Economic Studies 14,1

The

relationship between the monopoly and competitive

prices

posited

by

Losch has itself attracted

considerable attention, and is relevant to the work of Chamberlin. Greenhut et

al.

(1975) have

pointed out that the Loschian equilibrium will be characterised

by

higher prices than would spatial

monopoly, and that the more intuitively plausible outcome associated with traditional (non-spatial)

theory requires the explicit assumption that firms' prices at their respective market boundaries

be

fixed.

The Greenhut-Hwang-Ohta interpretation differs from that of Beckman

(1970),

who argues

that Loschian free entry

will

reduce prices to the minimum

level

consistent with

zero

profits. However,

as Capozza and

van

Order

(1977)

have pointed out, the debate hinges upon different interpretations

of "equilibrium". If

the

equilibrium condition requires profits to be

zero,

then Loschian competition

yields this outcome, but only at a "high" price; alternatively, if the condition requires all firms

to have equal market shares, then the Greenhut-Hwang-Ohta approach yields a more satisfactory

result. Capozza and van Order (1978) employ ZCV to show that Losch's perverse results obtain

only in exceptional cases, and not in the more traditional (Chamberlinian) framework of

monopolistic competition.

14.

The processes involved in product selection by firms have been the subject of study for some time.

Brems

(1951)

addressed the question but found it to be largely intractable. It

is

only since the work

of Baumol (1967) that potentially fruitful analysis has been possible. His model employs the

techniques of integer programming to establish optimal strategies of product design. On the

assumption that no retaliation by rivals is possible, he shows that the optimal strategy of a firm

involves making its product more, rather than

less,

similar to existing varieties. In contrast, Schuster

(1969) has considered a case in which the preferred strategy involves the production of varieties

which are quite distinct from those of rivals. Reichardt (1962) has cast the analysis in explicitly

game-theoretic

terms.

Lancaster

(1966)

has also studied the problem in a "characteristics" framework.

Valuable though these contributions have been, all are restricted to the analysis of product selection

in markets where N is fixed.

15.

For a diagrammatic exposition of some of the models set out here, see de Meza (1983).

16.

But see Pettengill (1979) for a counter argument.

References

Abbott, L.,

Quality

and

Competition,

New York, Columbia University Press, 1955.

Ali,

M.M. and Greenbaum, S.I., "A Spatial Model of the Banking Industry", Journal of

Finance,

Vol.

32,

1977, pp. 1283-1303.

Archibald, G.C., "Chamberlin versus Chicago", Review of Economic

Studies,

Vol. 29, 1961, pp. 2-28.

Archibald, G.C. and Rosenbluth, G., "The 'New' Theory of Consumer Demand and Monopolistic

Competition",

Quarterly

Journal of Economics, Vol. 89, 1975, pp. 569-590.

Archibald, G.C, Eaton, B.C. and Lipsey, R.G., "Address Models of Value Theory", in Stiglitz, J.E. and

Mathewson, G.F., (Eds.), New

Developments in the Analysis

of

Market

Structure,

London, Macmillan,

1986.

Bain, J.S., "Chamberlin's Impact on Microeconomic Theory", in Kuenne, R.E., (Ed.), Monopolistic

Competition

Theory:

Studies in

Impact,

New York, John Wiley and Sons, 1967.

Barzel, Y., "Excess Capacity in Monopolistic Competition",

Journal

of

Political

Economy,

Vol. 78, 1970,

pp.

1142-1149.

Baumol,

W.J.,

"Calculation of Optimal Product and Retailer

Characteristics:

the Abstract Product Approach",

Journal of

Political

Economy,

Vol. 75, 1967, pp. 674-685.

Baumol, W.J., "Contestable

Markets:

An Uprising in

the

Theory of Industrial

Structure",

American Economic

Review,

Vol. 72, 1982, pp. 1-15.

Baumol, W.J., Panzar, J.C. and Willig, R.D.,

Contestable Markets

and

the Theory

of

Industrial

Structure,

New York, Harcourt Brace, Jovanovich, 1982.

Beckman, M.J., "Equilibrium

vs.

Optimum: Spacing of Firms and Patterns of Market Areas",

Discussion

Paper 16, Brown University, 1970.

Bishop, R.L., "Elasticities, Cross-Elasticities, and Market Relationships",

American Economic

Review,

Vol.

42,

1952, pp. 781-803.

Chamberlin and Industrial Organisation Theory 51

Bishop, R.L., "Reply" American Economic Review, Vol. 43, 1953, pp. 916-924.

Bishop, R.L., "Reply", American Economic Review, Vol. 45, 1955, pp. 382-386.

Bishop, R.L., "Monopolistic Competition and Welfare Economics", in Kuenne, R.E., (Ed.), Monopolistic

Competition Theory: Studies in Impact, New York, John Wiley and Sons, 1967.

Brems, H.J., Product Equlibrium Under Monopolistic Competition, Cambridge, Harvard University Press,

1951.

Capozza, D.R. and van Order, R., "Pricing Under Spatial Competition and Spatial Monopoly, Econometrica,

Vol. 45, 1977, pp. 1329-1338.

Capozza, D.R. and van Order, R., "A Generalised Model of Spatial Competition", American Economic

Review, Vol. 68, 1978, pp. 896-908.

Capozza, D.R. and van Order, R., "Product Differentiation and the Consistency of Monopolistic Competition:

A Spatial Perspective", Journal of Industrial Economics, Vol. 31, 1982, pp. 27-39.

Carruthers, N., "Location Choice when Price is also a Decision Variable", The Annals of Regional Science,

Vol. 15, 1981, pp. 29-42.

Chamberlin, E.H., "Product Heterogeneity and Public Policy", American Economic Review, Vol. 40, 1950,

pp.

85-92.

Chamberlin, E.H., "Elasticities, Cross-Elasticities, and Market Relationships", American Economic Review,

Vol. 43, 1953, pp. 910-916.

Chamberlin, E.H., Towards a More General Theory of

Value,

New York, Oxford University Press, 1957.

Chamberlin, E.H., "The Origin and Early Development of Monopolistic Competition Theory", Quarterly

Journal of Economics, Vol. 75, 1961, pp. 515-543.

Chamberlin, E.H., The Theory of Monopolistic Competition, Cambridge, Harvard University Press, 8th

edit., 1962.

Copeland, M.A., "Competing Products and Monopolistic Competition", Quarterly Journal of Economics,

Vol. 55, 1950, pp. 1-35.

Cournot, A.A., Researches into the Mathematical Principles of the Theory of Wealth, New York, A.M.

Kelley, 1938.

D'Aspremont, C, Gabszewicz, J-J and Thisse, J.F., "On Hotelling's Stability in Competition", Econometrica,

Vol. 47, 1979, pp. 1145-1150.

Demsetz, H., "The Nature of Equilibrium in Monopolistic Competition", Journal of Political Economy,

Vol. 67, 1959, pp. 21-30.

Demsetz, H., "The Welfare Implications of Monopolistic Competition", Economic Journal, Vol. 74, 1964,

pp.

623-641.

Demsetz, H., "Do Competition and Monopolistic Competition Differ?" Journal of Political Economy,

Vol. 76, 1968, pp. 146-148.

Demsetz, H., "The Inconsistencies of Monopolistic Competition: A Reply", Journal of Political Economy,

Vol. 80, 1972, pp. 592-597.

Devletoglou, N.E., "A Dissenting View of Duopoly and Spatial Competition", Economica, Vol. 32, 1965,

pp.

140-160.

Dewey, D., "Monopolistic Competition No Bar to Efficient Production", Journal of Political Economy,

Vol. 66, 1958, pp.

24-33.

Dixit, A.K., "A Model of Duopoly Suggesting a Theory of Entry Barriers', Bell Journal, Vol. 10, 1979,

pp.

20-32.

Dixit, A.K. and Stiglitz, J.E., "Monopolistic Competition and Optimum Product Diversity", American

Economic Review, Vol. 67, 1977, pp. 297-308.

Eaton, B.C., "Free Entry in One-Dimensional Models: Pure Profits and Multiple Equilibria", Journal of

Regional Science, Vol. 16, 1976, pp. 21-34.

Eaton, B.C. and Lipsey, R., "The Principle of Minimum Differentiation Reconsidered: Some New

Developments in the Theory of Spatial Competition", Review of Economic Studies, Vol. 42, 1975, pp.

27-49.

Eaton, B.C. and Lipsey, R., "The Non-uniqueness of Equilibrium in the Loschian Model", American

Economic Review, Vol. 66, 1976, pp. 7-93.

Eaton, B.C. and Lipsey, R., "Freedom of Entry and the Existence of Pure Profit", Economic Journal, Vol.

88,

1978, pp. 455-469.

52 Journal of Economic Studies 14,1

Economides, N., "The Principle of Minimum Differentiation Revisited", Discussion Paper, 134, Columbia

University, 1982.

Fama, E. and Laffer, A., "The Number of Firms and Competition", American Economic Review, Vol. 62,

1972,

pp. 670-674.

Fellner, W., "Elasticities, Cross-Elasticities and Market Relationships: Comment", American Economic

Review, Vol. 43, 1953, pp. 898-910.

Fellner, W., "The Adaptability and Lasting Significance of the Chamberlinian Contribution", in Kuenne,

R.E., (Ed.), Monopolistic Competition Theory: Studies in Impact, New York, John Wiley and Sons, 1967.

Formby, J.P. and Smith, W.J., "The Chamberlin and Stackelberg Duopoly Relation", Economic

Record,

Vol. 55, 1979, pp. 368-370.

Friedman, J.W., Oligopoly and the Theory of Games, Amsterdam, North Holland, 1977.

Friedman, J.W., Oligopoly Theory, New York, Cambridge University Press, 1983.

Gabszewicz, J.J., "Blue and Red Cars, or Blue Cars Only? A Note on Product Variety", Economica, Vol.

50,

1983, pp. 203-206.

Gabszewicz, J.J. and Thisse, J.F., "Entry (and Exit) in a Differentiated Industry", Journal of Economic

Theory, Vol. 22, 1980, pp. 327-338.

Gal-Or, E., "Hotelling's Spatial Competition as a Model of

Sales",

Economics Letters, Vol. 9, 1982, pp. 1-6.

Gannon, C.A., "Consumer Demand, Conjectural Interdependence, and Location Equilibria in Simple Spatial

Duopoly", Regional Science Association Papers, Vol. 28, 1972, pp. 83-107.

Grace, S.H. jr., "Professor Samuelson on Free Enterprise and Economic Efficiency", Quarterly Journal

of Economics, Vol. 84, 1970, pp. 337-340.

Graitson, D., "On Spatial Competition Processes", Working Paper, Université de Liege, 1979.

Graitson, D., "On Hotelling's Stability in Competition, Again", Economics Letters, Vol. 6, 1980, pp. 1-6.

Greenhut, M., Hwang, M. and Ohta, H., "Observations on the Shape and Relevance of the Spatial Demand

Function", Econometrica, Vol. 43, 1975, pp. 669-682.

Hart, O.D., "Monopolistic Competition in a Large Economy with Differentiated Commodities", Review

of Economic Studies, Vol. 46, 1978, pp. 1-30.

Hart, O.D., "Perfect Competition and Optimal Produc</