On Equilibrium in Monopolistic Competition

Author(s): Richard Carson

Source: Eastern Economic Journal , Summer, 2006, Vol. 32, No. 3 (Summer, 2006), pp.
421-435
Published by: Palgrave Macmillan Journals

Stable URL: https://www.jstor.org/stable/40326288

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 [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Palgrave Macmillan Journals is collaborating with JSTOR to digitize, preserve and extend access
to Eastern Economic Journal

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC
COMPETITION

Richard Carson
Carleton University

INTRODUCTION

This paper is about equilibrium under monopolistic competition,

the idea that each seller in such a market must have unique, produ
inputs whose uniqueness allows it to earn rent, even in long-run eq
existence of this rent affects our interpretation of equilibrium in a fun
Monopolistic competition requires specialized inputs because some prod
tiation is compatible with perfect competition [Rosen, 1974]. If we thin
service as a bundle of attributes in the manner of Lancaster [1966, 1971
ent product could be a different combination of the same attributes. Pe
tion in the supply of each attribute could then result in perfect comp
supply of products. Firms would be price takers, even though no two
the same good or service. It is when each firm imparts a unique attrib
put - one not exactly duplicated by any other supplier and therefore on
perfect substitute - that we leave the world of perfect competition, bo
and in products.
In order to supply an attribute that no competitor is able to provide
would have to be protected by a barrier that gives it a cost advantage in
attribute, or else the advantage would have to come from possession o
indivisible input that is specialized to this attribute, and therefore to t
uct. Since there are no entry barriers under monopolistic competit
must be the sole possessor of one or more specialized inputs. Without
specialized inputs, it is hard to explain why monopolistic rather than pe
tion prevails.
In this paper, customers are assumed to demand goods, but we can st
differentiated product as possessing unique attributes, features, or pr
cost-effective supply requires inputs specialized to these unique elemen
the product. For simplicity, to "differentiate" a product will mean to im
attribute to it. For example, suppose that each bakery in a city supplie
a distinctive flavor, which can be varied within limits without destroyi
ity. When this unique element is present, every bakery faces downwa
mand. Its product-specialized inputs are its bread formula or recipe plus
tacit knowledge required to create the bread from the formula at low
sive of rent. These inputs potentially earn rent that cannot be compet

Richard Carson: Department of Economics, Carleton University, C870 Loeb Buil

By Drive, Ottawa, Canada, K1S 5B6. E-mail: [email protected].

Eastern Economic Journal, Vol. 32, No. 3, Summer 2006

421

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
422 EASTERN ECONOMIC JOURNAL

When product-specialized inputs are mobile between fir

the opportunity costs of the suppliers for which th
Chamberlinian equilibrium [Chamberlin, 1933]. There eac
is tangent to its average cost where both slope downward
that minimizes average cost. Chamberlin therefore co
competitor would choose production facilities that are be
under increasing rather than constant returns to sca
facilities with excess capacity. However, it would be mis
tion facility is below optimal size or has excess capacity
downward solely because of the inclusion of rent in cost.
input, which in a sense has "excess" capacity. For examp
ies can also supply a non-differentiated or generic bread
uct-specialized input. The generic bread would be supplie
but switching from the generic to the differentiated bre
long-run equilibrium output at all.
Thus we must be careful to distinguish between cost
on product-specialized inputs and cost measured exclusiv
derstanding, our basic results are as follows. First, the p
monopolistic competitor are found by maximizing the di
and its cost measured exclusive of the rent on its pro
maximized difference equals the rent in question. The in
gives rise to the traditional Chamberlinian solution, in w
cost is tangent to demand and therefore downward-slop
average cost will lie below demand in long-run equilibriu
positive - and may be constant or even upward-sloping.
gency between demand and rent-inclusive average cos
quality, or output; rather, tangency occurs at the price,
the rent reaches its maximum.
Suppose we measure economies of scale using rent-exclusive average cost. Then,
under an additional assumption to be spelled out and motivated below, a monopolisti
competitor's rent-exclusive average cost will be downward-sloping, constant, or up-
ward-sloping in long-run equilibrium, depending entirely on whether differentiatin
its product increases, leaves constant, or decreases returns to scale in production and
supply. This is because differentiating its product does not change its equilibriu
output from the case in which it would supply a non-differentiated version of its go
or service under perfect competition. Thus when economies of scale in supplying a
differentiated product are no greater than those in supplying the non-differentiate
version, a monopolistic competitor produces where its rent-exclusive average cost is
constant or upward-sloping. Its Lerner index of market power can then be no greater
than the equilibrium share of rent in its value added. The smaller is this share - and
in this sense, the more intense is the competition facing it - the more closely will its
equilibrium approximate that of a perfectly-competitive firm, although at the cost o
suppressing the unique element in its product differentiation.
More generally, a monopolistic competitor will set socially optimal quality levels
for types of quality change that play no role in product differentiation. However, wh

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 423

quality change affects the slope of its demand curve, the firm will set a level o
that leaves this slope flatter than social optimization would require. In thi
firms do not differentiate their products enough. Thus the two sources of inef
in long-run equilibrium are the traditional tendency to price above marginal co
the tendency of each firm to under-differentiate its product. Under the add
assumption mentioned above, free entry and exit produces too many firms, th
number, or too few, depending on whether the tendency for greater compet
reduce product differentiation by already-existing firms has a greater, equal, o
impact on welfare than the increase in product diversity brought about by the
ance of a new product. Without the tendency to under-differentiate, entry w
socially efficient.
Finally, suppose product-specialized inputs are attached to specific firms,
than mobile between them. Then rent is not included in opportunity cost, and
will earn positive profit in long-run equilibrium. Because of free entry and e
profit will equal the rent on its specialized inputs. It will maximize its profit,
to say that it will find a price, output, and quality combination that maximiz
rent. Provided every firm produces efficiently, the industry will therefore re
same long-run equilibrium, with respect to outputs, prices, and quality levels,
product-specialized inputs are mobile between firms.1 If a specialized input is
firms compete for this input, and the winner must pay the maximized rent in
bid successfully for the input. Rent is therefore included in opportunity cost.
case, a monopolistic competitor must set price, quality, and output to maxim
difference between its revenue and its cost measured exclusive of rent, and this
ence is the rent on its specialized inputs.

THE FIRM IN MONOPOLISTIC COMPETITION

In this paper, firms are allowed to vary quality as well as price and
monopolistic competition with quality variation is defined by three basic
[Chamberlin, 1933; Dorfman and Steiner, 1954] . (a). Owing to product di
between firms, each firm faces downward-sloping demand when quality
stant, (b). In the long-run, the industry reaches a Nash equilibrium [Na
is self-enforcing, in the sense that no one firm acting on its own can incr
by moving away from it. As a result, we can treat each firm as setting pr
independently of the others, (c). Because of free entry and exit, includ
mobility of inputs between firms, each monopolistic competitor maxim
zero in long-run equilibrium. This leads to the Chamberlinian outcome, w
erties noted above, which has been controversial [eg., Demsetz, 1959, 19
1985, 1989]. I shall outline some of the historical controversy below
relation to the present paper, although in order to do this, I must first s
model.
To this end, I shall use capital letters - X, Y, andZ - to designate speci
and small letters to denote output quantities of these products. Suppose
product monopolistic competitor sells x units of a differentiated produ
assumption that the income effects of its output and quality changes are

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
424 EASTERN ECONOMIC JOURNAL

to ignore. Let P(x,q,r) be the total value of these x

amount paid for them plus the consumer surplus - w
ous function of x, as well as of q, a quality index tha
firm's product, and r, a quality index that plays no r
More precisely, changes in q affect the slope of the f
changes in r do not. The partial derivative of P with
second-order continuous demand price, Px = F(x,q,r), a
+ g(r). If subscripts denote partial derivatives, the cros
Let x*, g*, and r* be the long-run equilibrium valu
the firm finds it profitable to differentiate its prod
unique attribute to it, fjx,q) < 0 for values of q in som
rium. However, for another value of q - say q = q0 - t
product, and without loss of generality, we can take
that this firm has just one product-specialized input.
input adds no value to its good or service, and the firm
tition. However, the specialized input does add value i
increases in q also add value to the bread (or shift
therefore <?* > 0. Product differentiation is assum
increases from 0. That is, as q tends to zero, so doe
unique attribute to its good or service, its quality inde
attribute. In our bakery example, if two firms set q = 0, b
when each sets q = l9 their breads taste differently, since
In general, Px will depend on price-quality combinat
well as onx, q, and r. To reach a Nash equilibrium,
profit independently of one another. This obliges the
prices, outputs, and quality levels of competitors, whi
rium. Given these expectations, each firm's demand pr
and quality.
Besides its product-specialized resource, the firm will use non-specialized inputs,
which are assumed to be in horizontal supply to it (although possibly in upward-slop-
ing supply to the industry). Thus in addition to the rent on its specialized input, sup-
pose the firm incurs a cost of K(x,q,r) to differentiate its output. If C(x,q,r) is the rent-
exclusive cost of x units of output with quality levels q and r, K(x,q,r) = C(x,q,r) -
C(x,0,r), where C(x,0,r) is the cost of x units of the generic substitute for X- call it Y-
with quality level r, supplied under perfect competition. The average cost of Y, ACy =
C(x,0,r)/x, is assumed to have the traditional U-shape, with a unique minimum. With-
out the specialized input, the value of x units of output would be Pyx, where Py is the
price of the generic substitute, and the contribution of the product-specialized input to
value added is therefore (Px - P)x- K(x,q,r).

EQUILIBRIUM QUALITY

The rent on the firm's product-specialized input is the maximum value of

k = Px - C(x,q,r) over q, r, and x. The first-order conditions for maximizing n
with respect to q and r are

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 425

(1) Pxx = Cr = MCr,

(2) Pxqx = Cq = MCq,

where Pxr = g (r) and P -f are the partial derivatives of

and Cr = MCr and C =MC are the partial derivatives ofC
indices, or the marginal costs of r and q. In order to ev
Pr and Pq be the increases in P generated by unit increas
values to buyers of r and q. Then the socially optimal lev
= MCr and P = MC . When jc = 0, Pr = JP =0. Using the
therefore write Pr(x) = P\x, where P*xr is an intermediat
and jc. However, Pxr = g (r) is independent of x9 so that
therefore becomes

(3) Pr = MCr.

In long-run equilibrium, the firm se

The same is not true of q, howeve
We again have Pq(x) = P^x, but P\q
consider the special case in which P
q = 0, small increases in q must red
becomes negative. Therefore P =
steeper. As shown in Figure 1 below,
by a unit increase in q, is then rela
vertically-shaded area, ABCE, is P
AFCE, is Pq - P1^. The firm ignores
setting q because it does not contrib
consumer surplus.
Because P <
xqx Xi/ xq n
0, we have P1 > P n

(4) Pq>Pxqx = MCq,

and the profit-maximizing level of q is below the so
that purely infra-marginal shifts of demand do not a
standpoint of revenue change, the firm sees every q
a parallel shift by an amount equal to the resulting
an upward shift in demand caused by an increase of
parallel shift followed by a tilt. In Figure 1, the tilt
consumer surplus, which has no effect on the firm's
in setting q. Similarly, an increase of q that makes
general, possible at values of q above 0 - amounts to
downward that lowers consumer surplus. Again the
such an increase may therefore be profitable when

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
426 EASTERN ECONOMIC JOURNAL

FIGURE 1

The net result is a tendency to set q where product differentiation - as m

by the slope of the demand curve - is below the socially-optimal level. When
in q reduce product differentiation (or make the demand curve flatter), they
valued by the firm, in comparison to its customers. When increases in q rais
differentiation, they are under-valued by the firm, in comparison to its cust
particular, around q = 0, we always have P^ < 0, and therefore P > Px x. Some
supplying Y might well differentiate their outputs by imparting unique att
them if they were able to capture all of P , and the ability to internalize all o
even cause Y to vanish from the market.
Finally, the increase of q from 0 to q* creates both consumer and producer sur-
plus. It creates the former by differentiating the product, which makes the firm's
demand curve steeper. By comparison, a single perfectly competitive firm faces hor
zontal demand and therefore generates negligible consumer surplus. Similarly, t
increase of q creates producer surplus in the form of rent to the product-specialize
input, which rent can not be realized if these inputs are used to produce the generi
alternative. This increase is therefore welfare-improving, even though q* is not the
socially optimal level of q, and product differentiation causes output to be set where
exceeds the marginal cost of x.

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 427

EQUILIBRIUM QUANTITY

In fact, the familiar first-order condition for maximizing n = Px - C(x,q,r) with

respect to x is

(5) MRx = MCx,

where MRx = PX + P^x and MCx = Cx are t

Because P^ < 0, too little of X is supplied,
inputs were transferred from the generic
;r* be the firm's total cost including rent
note that MCq, MCr, and MCx are also th
Thus as long as the firm is viable, equatio
for maximizing either Px - C(x,q,r), whic
puts are immobile between firms, or Px -
specialized inputs are mobile between fi
equations (1), (2), and (5) must hold, along

(5a) px=AC*x>

where ACRx - CRlx is the firm's rent

where ACRxx is the slope of ACRx. AC
as long as product differentiation is
ACRx are increasing.
However, let AC x = C(x,q,r)/x be t
ACx could still be constant or upward
scale as measured by AC x would be c
falling solely because tP is included i
the firm's own-price elasticity of de
Pjc denote this elasticity. Since AC^ =

(6) ep > Pjc/ ti*,

in long-run equilibrium. The firm's elasticity o

to the reciprocal of the share of rent in its va
turns to scale, as measured by ACx, prevail. Th
least 5; if this share is only 10%, ep must be a
these elasticities to be on the high side, althou
between one and infinity.
In order to investigate this issue further, w
that will allow us to locate the equilibrium level
earlier literature on the question of excess cap
As noted above, Chamberlin argued that long-r
excess capacity and production facilities below

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
428 EASTERN ECONOMIC JOURNAL

the envelope theorem plus the fact that ACRx is downw

AC*^ < 0. Writing a quarter of a century later, however,
this excess capacity result. To understand his argume
profit-maximizing levels of q and r at each given x, and
by Px(x) = Px(x,q(x),r(x)) and ACRx(x) =ACRx(x,q(x),r(x)l Th
tangent where each has a slope of zero, since increases in
and/or r that would raise Px and offset the negative dir
increases in x.2 In this case, Demsetz claimed, no exces
And while capacity should be measured for a specific and
would appear to mean for a fixed q and r - Demsetz argu
could represent higher selling costs, rather than increas
over, economies of scale in selling costs could cause q(x) a
functions of x over some ranges of output [Demsetz, 195
Demsetz' article aroused interest in the subject, but w
[1970] and Schmalensee [1972], among others. The thr
that if buyers are not misled, increases in q and/or r
broadly-defined quality for at least some customers. (On t
[1985, 266-270] summary.) Demsetz [1972] accepted this c
entirely new argument to the effect that, in comparing m
tition, CRx is an inappropriate measure of cost. This is be
rather than a "non-branded" version of a product - analo
than Y above - saves the buyer certain costs.3 Demsetz su
CRx and claimed that the resulting measure of average co
demand and ACRx are tangent. However, if X were to dis
unclear whether replacing units of X with units of Y and
be a utility-maximizing strategy for buyers. Instead, they
adjusting their consumption bundles. In place of these co
ence of a product, Z, introduced below, which will anchor
rate of output.
Subsequently, Barzel and Schmalensee were criticized b
and Murphy [1978] tried to re-establish Demsetz' origina
his 1985 survey article, Margolis maintained that efforts by O
[1974] to overturn Chamberlin's excess capacity theorem w
himself argued [1989] that a multi-product monopolis
brand name does not necessarily operate under excess ca
spill over from one product [sold under this brand name]
challenges to the excess capacity theorem, this one appea
but it relies on the existence of benefits that are exte
although internal to the firm supplying these products.
which can exist for multi-product firms only - Chamber
a single-product firm) have fared better than one might
In general, this literature did not deal with product-sp
on these and therefore did not distinguish between rent
cost. Its claims and counter-claims relate to rent-inclusive
ACRx as downward-sloping at the equilibrium value of x

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 429

while maintaining that this may be due entirely to the existence of rent on p
specialized inputs and therefore misleading. The goal here is to find condition
which rent-exclusive average cost, or ACx, will be minimized or upward-slopi
equilibrium, in the sense that AC^ > 0.
With this in mind, we note that (Px - P ) > (ACRx - AC ) must hold in
run equilibrium for the firm supplying X, where ACy - C(x,0,r)/x. Sin
ACRx, (Px - Py) < (ACRx-ACy) would imply that Py > Ac"y is possible and the
that positive economic profit could be earned in supplying Y. The additional a
tion mentioned above is then that at least one product, Z, is available, with p
which meets three conditions. First, Z is supplied under free entry and exit, w
result that Pz = ACRz in equilibrium, where ACRz is the rent-inclusive average
Z. In particular, Z may be supplied under perfect or monopolistic competition.
Z can be used together with Y, and this combination competes with X in the
Z is therefore assumed to be an imperfect substitute for X, and the same is t
other products that compete in the same industry with Z. An expansion of the
of these substitutes which puts downward pressure on Pz is also assumed to pu
ward pressure on Px.
Third, a supplier of Z is at least as cost effective as the supplier of Xin addin

to Y. If the firm supplying X would supply Y instead, its average cost would fal
Increasing q from 0 to q* raises the firm's price from Py to Px and its rent-in
average cost from ACy to ACRx. Thus suppose the supplier of X is producing an
at which ACRx is downward-sloping and that demand is such that Px = ACRx. T
meet the third condition, a supplier of Z must be able to produce an output of
at which

(7) ACRzi(ACRx-ACy\
when output units of Z are scaled in such a way that Pz =
bread example, suppose an entrepreneur markets a spread,
of the generic bread, Y, but clashes with the taste of the d
result, the combination of generic bread with spread compe
bread.
When the third condition is met and the Y-industry is
supplier of X must produce on a scale that is efficient for
with the composite good whose units consist of a unit of Z
average cost equals (ACRz + P). Any other output puts X at
suppose x is set where Px = ACRx and (Px - Py) > (ACRx -ACy) o
But then long-run equilibrium can not prevail, since equatio
is possible for any product, Z, meeting the above three cond
prospect of quasi-rents will attract entry of new competitor
Z competes, and this will put downward pressure on Pz an
ACRx or ACRz, since this entry would be foreseen. By the

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
430 EASTERN ECONOMIC JOURNAL

competitive pressures on X will have increased, andP^

of X will have to change its output or exit the market
Thus if this firm survives over the long run, it m
(ACRx-ACy). If MCy is the marginal cost of Y, x* is th

(8) Py = MCy=ACy,
or at which AC = C(x,09r)/x reaches its minimum
conditions hold, differentiating a monopolistic com
its equilibrium output, since a firm supplying Y als
minimum. In addition, this is the only output at whic
specialized input equals this input's rent- that is at which
If there is no product with the properties of Z, l
where (Px-P)> (ACRx-AC ). But then there is an in
such a product, which will be viable over the long ru
run, as long as (Px - P ) > (ACRx - AC ), which impli
assumptions behind equation (7). If Z is also supplied
with its own generic alternative, W, the above analys
and X. That is, if we reverse the roles of X and Z an
third condition then holds - z* will be where AC w r
Finally, while the existence of Z allows us to ancho
equilibrium output where AC is minimized, this e
result that equilibrium under monopolistic compe
rent-exclusive average cost is minimized or upward-
on the firm's specialized input may be maximized in

FURTHER RESULTS

For the remainder of this paper, we measure returns to scale using

exclusive average cost of the monopolistic competitor supplying X
when equation (7) holds, differentiating this firm's product does not ch
rium output. Thus it produces where returns to scale are increasin
decreasing, depending entirely on whether differentiating its product
constant, or decreases returns to scale in production and supply at the
minimizes AC . In particular, if product differentiation does not alter
these will be the same for X as they are for Y at any given output. A
both reach their minima at x = ¿c*, where we consequently have

(9) ACx = MCx.

Constant returns to scale in the suppl

2 below, which is drawn with q and r fi
AC*
is tangent
y x
tox
the x
horizontal
y
line,
margin
where i

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 431

specialized input. However, it should be stressed that equation (9) hold

product differentiation does not alter returns to scale, whereas equat
whenever inequality (7) does.

FIGURE 2

More generally, suppose that economies of scale in supplying the differ

product, X, are less than or equal to those in supplying the non-differentiat
Y. In long-run equilibrium, the firm supplying X will then produce where A
stant or upward-sloping (AC^ > 0), and it must be earning positive rent, in
ACRx and ACx are clearly different. Because MCx > ACx also holds, inequalit
floor on the firm's own-price elasticity of demand equal to the reciprocal of
of rent in its value added. Its Lerner index of market power (= (Px - MCx)/P
the same as 1/e in equilibrium, can be no greater than this share (or no gr
tPIPx). Thus the more intense is the competition facing a monopolistic com
in the sense that the greater is the pressure on its rent - the smaller will b
of market power and the more closely will its equilibrium approximate th
fectly-competitive firm.
The latter outcome is assured by the fact that, regardless of rent, the m
competitor produces where ACy = C(x,0,r)/x reaches its minimum, which is
demand is tangent to ACRx. As rent disappears, ACRx tends to ACx, but p

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
432 EASTERN ECONOMIC JOURNAL

never occurs where AC x is downward-sloping, and demand

if income effects are small enough to ignore. Thus the limi
horizontal, but this implies that the unique attribute in pr
disappeared and that q* has tended to zero. As a result, the
in Figure 2, where AC is minimized and equal to both P an
tends to zero, but the unique attribute in product differen
this is, in general, not consistent with efficiency, owing to
capture all of P .
Only if product differentiation increases returns to scale
produce where AC x is downward-sloping, when equation (
e.g., if differentiating the product increases the ratio of fixed
at x*.5 Then it is possible that the smaller is t& /Pxx, the m
equilibrium approach a limiting outcome where q* is positi
are tangent and downward-sloping. However, such a sol
product-specialized inputs earn no rent and scale economies
tiated product are greater than those in supplying its gener
outcome has the advantage of preserving the unique eleme
differentiation, although it has the disadvantage that pric
In order to consider the welfare effects of entry of new com
try, we return to the more natural assumption that differen
it a unique attribute reduces returns to scale or leaves them
that a new entrant marginally reduces the demand for
previously existing supplier). If this would cause an output
consumers would be roughly PxAx, whereas the cost savin
welfare loss of (Px - MCJAx that the entrant does not take
entry decision. This is sometimes called the "business-stealin
a new product, which in of itself increases product variety
changes in products already supplied. The net impact of th
welfare - the "product-diversity" effect - that the entrant
tomers value variety, the number of competitors will b
optimal number, depending on whether these business-ste
effects produce a net decrease or increase in welfare [Mank
However, there is no business-stealing effect for Y or for
marginal cost (including r), nor is there any for products
stant in the face of entry.
Therefore, would entry cause the equilibrium value of x
there are no long-run fixed or set-up costs in supplying Y,
change relative input prices. In either case, the output tha
an existing supplier will be invariant to entry of new com
firm is forced out of this market, entry would not change the
long as inequality (7) holds before and after entry. And whil
exit the industry, these are most likely to be marginal supp
near marginal cost, especially since firms operating under
are small relative to industry size. Entry could steal busine
industry - and/or cause the supply of complementary good

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 433

welfare gain or loss resulting from this is small, the same will be true of the b
stealing effect.
As a result, entry would then be socially efficient if suppliers were able to
all of Pq - or if q were fixed, as in the classic paper by Dixit and Stiglitz [1977]
cient entry arises only because in long-run equilibrium, P ^ MC . A given
may cause q to rise or fall, but as rents are squeezed by greater competition,
a tendency for q to tend to zero, and therefore for the unique element in pr
differentiation to decline and eventually vanish. Thus whether and how entry
this industry departs from the socially optimal level depends on whether and h
tendency for greater competition to reduce product differentiation by already-
firms has a greater or lesser impact on welfare than the increase in product di
brought about by the appearance of a new product with a new unique attribu
ever, if entry changes relative input prices and there are positive fixed or set-
in supplying Y, entry could alter the ratio of fixed to variable cost at any outp
thereby change the output at which C(x,0,r)/x reaches its minimum. This outp
either rise or fall, but the resulting effect on welfare would not be taken into
by an entrant, and the tendency to produce where Px > MCx would also be a so
inefficient entry.

CONCLUSION

To sum up, the price, output, and quality of a monopolistic competito

mined by maximizing the difference between its revenue and its cost, w
measured exclusive of the rent on its product-specialized inputs. Such a f
have unique inputs that are specialized to its unique product - since produ
tiation is otherwise compatible with perfect competition - and the uniquen
inputs allows them to earn positive rent, even in long-run equilibrium. T
of rent in cost gives rise to the traditional Chamberlinian solution, in wh
inclusive) average cost is tangent to demand and therefore downward-slop
rent is excluded, average cost may be constant or even upward-sloping in e
and in this sense, monopolistic competition need not give rise to excess c
production facilities that are too small.
Perhaps the point to emphasize in closing is that differentiating a mo
competitor's product by increasing q from 0 to q* creates both consumer a
surplus. The increase of q therefore improves welfare, even though q
socially optimal level of q, and product differentiation prevents marginal-
of output. To restore marginal-cost pricing by suppressing this product dif
would reduce welfare and, in particular, would reduce the productivity of
specialized inputs by destroying the consumer and producer surplus that
This surplus can only be realized when firms are allowed to differentiate
ucts, not when they supply non-differentiated or generic alternatives.
In addition, when equation (7) holds, differentiating a firm's product do
its equilibrium output. If economies of scale, as measured by rent-exclusi
no greater in supplying the differentiated product than those in supplyi
version, a monopolistic competitor produces where AC x is minimized or upw

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
434 EASTERN ECONOMIC JOURNAL

in the sense that AC^ ^ 0. In this case, there is also no ex

capacity of the product-specialized input itself, and retu
ACx - although not as measured by ACRx - are constant
ventional criticisms of monopolistic competition are often
ignore the most important thing about it, namely that
rational and knowledgeable, it improves welfare by creat
sumer and producer surplus that goes with this - without
at all. The story these criticisms tell, as Schumpeter [
related context, "is like Hamlet without the Danish princ

NOTES

I wish to thank Erwei Yin, Han Li, and two anonymous referees for assistance or h
ments. Sole responsibility for errors rests with me. An earlier version of this paper
at the Southwestern Economic Association meetings in Corpus Christi on March 17,
1. An industry operating under monopolistic competition may have multiple equilibria
the equilibria will be the same regardless of whether product-specialized inputs a
tween firms, provided all firms produce efficiently.
2. In terms of earlier notation, the slope of Px(x) is P^ + Pxqq'(x) + Pxf'(x), where q'(x) an
derivatives of q{x) and r(x) with respect to x. We have seen that both P^ and Pxr are po
q and r are positive (equations (1) and (2) above), and q'{x) and/or r'(x) could be posit
scale economies in supplying q and/or r. Thus P^ + P^q'ix) + Pxrr'(x) could be non-n
Px(x\ q(x\ and r(x) are the profit-maximizing values of Px, q, and r at each given x, Px
as Demsetz' [1959] mutatis mutandis average revenue curve, MAR.
3. Demsetz [1972] only sketches his argument, but his basic idea is that, in comparing
with perfect competition, CRx is an inappropriate measure of cost. As noted below in t
is because consuming a "branded" rather than a "non-branded" version of a product
consuming X rather than Y above - saves the buyer certain costs. These are "cost
need to be incurred to ascertain the quality of the product, to establish prestigious con
some way other than by consuming branded commodities, and to be confident of c
responsibility should the product be defective in some respect. The consumer can r
costs by purchasing differentiated products, since product homogeneity makes it m
both to discern clear lines of responsibility for product quality and to consume con
[595-596] However, if X were to disappear from the market, it is unclear whether r
of X with an equal number of units of Y and incurring these costs would be a utilit
strategy for buyers. Instead, they might be better off simply adjusting their consump
Moreover, it is not obvious that product homogeneity must make it more difficult to d
lines of responsibility for product quality. Instead of these costs, I assume the exis
product, Z, introduced below.
4. Margolis [1989, 199].
5 Write C(x,q,r) as C = B(q,r) + V(x,q,r), where B is fixed cost - or cost that is independent of
output - and V(x,q,r) is cost that varies with x. We then have MCx = Vx, the partial derivative of V
with respect to x. The returns to scale embedded in C - or the firm's elasticity of production - are
given by S = ACJMCx = ClxCx = (B + V)/xVx = (R + 1)SV, where R = Bl V is the ratio of fixed to
variable cost, and Sv = V/xVx are the returns to scale embedded in V. Thus if increases in q raise
(1 + R) or Sy, they will also raise S, provided they do not reduce the other of these two components
by a larger percentage amount. Intuitively, however, it seems more natural to assume that
differentiating a product by giving it a unique attribute would reduce economies of scale, or at least
not increase them, although all three outcomes are possible.

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms
 ON EQUILIBRIUM IN MONOPOLISTIC COMPETITION 435

REFERENCES

Barzel, Y. Excess Capacity in Monopolistic Competition. Journal of Political Econ

October 1970, 1142-1149.
Chamberlin, E. The Theory of Monopolistic Competition. Cambridge, MA: Harvard
1933.

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

February 1959, 21-30.

May/June 1972, 592-597.

Dixit, A. K. and Stiglitz, J. E. Monopolistic Competition
Economic Review, June 1977, 297-308.
Dorfman, R. and Steiner, P. O. Optimal Advertising and Optimal Quality. American Economic
Rpvìpii) Dprpmher 1954. 826-836.

Greenhut, M. C. A Theory of the Firm in Economic Space. Austin: University of

Hotelling, H. Stability in Competition. Economic Journal, March 1929, 41-57.
Lancaster, K. A New Approach to Consumer Theory. Journal of Political Economy, April, 1966, 132-
159.

Mankiw, N. G. and Whinston, M. D. Free entry and

Spring 1986, 48-58.
Margolis, S. The excess Capacity Controversy: A C
April 1985, 265-275.

1989. 199-209.

Murphy, M. The Consistency of Perfect and Monopolistic Competition. Economic In

1978. 108-112.
Nash, J. Non-Cooperative Games. Annals of Mathematics, July 1951, 286-295.
Ohta, H. On the Excess Capacity Controversy. Economic Inquiry, April 1977, 153-165.
Rosen, S. Hedonic Prices and Implicit Markets: Product Differentiation in Pure Competition. J
of Political Economy, January/February 1974, pp. 34-55.
Schmalensee, R. A Note on Monopolistic Competition and Excess Capacity. Journal of Pol
Economy, May/June 1972, 586-591.
Schumpeter, J. A. Capitalism, Socialism, and Democracy. New York: Harper and Row, 195
Smithies, A. Optimum Location in Spatial Competition. Journal of Political Economy, Jun
423-439.

Spence, A. M. Monopoly, quality, and regulation. Bell Journal of Economics, Autumn 1975, 4

This content downloaded from

202.43.95.117 on Mon, 05 Oct 2020 12:18:51 UTC
All use subject to https://about.jstor.org/terms