European Economic Association

General Equilibrium with Imperfect Competition

Author(s): Egbert Dierker and Hildegard Dierker
Source: Journal of the European Economic Association, Vol. 4, No. 2/3, Papers and
Proceedings of the Twentieth Annual Congress of the European Economic Association (Apr.
- May, 2006), pp. 436-445
Published by: Wiley on behalf of European Economic Association
Stable URL: http://www.jstor.org/stable/40005109
Accessed: 12-05-2016 14:38 UTC

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

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Wiley, European Economic Association are collaborating with JSTOR to digitize, preserve and extend
access to Journal of the European Economic Association

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
 GENERAL EQUILIBRIUM WITH IMPERFECT
COMPETITION

Egbert Dierker Hildegard Dierker

Institut fur Hohere Studien and Institut Institut fur Hohere Studien and Institut
fur Volkswirtschaftslehre, Universitat fur Volkswirtschaftslehre, Universitat
Wien Wien

Abstract
The objective of a firm is not well-defined if firms have market power. We present an example
of Cournot-Walras competition in order to shed light on this problem and to motivate the
concept of real wealth maximization that B. Grodal and E. Dierker have proposed as the firm's
objective. (JEL: (JEL:
(JEL: D21,
(JEL: D21, D42, D42,
D21, D21, D42,D42,
D43, D43,
D43, L13,
D43, L13,L13,
L13,L21)
L21) L21)
L21)

1. Introduction

In Birgit Grodal's view, General Equilibrium (GE) Theory relies on the following
principles: All consequences of any economic decision must be fully taken into
account. Thus, a GE model must be closed. In a GE model all concepts must have
an unambiguous, well-defined meaning.
GE Theory is not based on the assumption of perfect competition. One of
Grodal's major goals over many years has been to improve our understanding of
GE Theory with firms that possess market power or with firms that act within an
incomplete system of markets. The central question is: What is the objective of a
firm?

The standard answer that firms should maximize profits is discussed in

Kreps (1990, p. 727) as follows: "The notion that shareholders necessarily want
managers to maximize profits is a long-standing component of the folklore of
capitalism. But it is incorrect folklore, if the firm has market power and if its
shareholders participate, even indirectly, in the markets that are affected by the
operations of the firm." Kreps adds: "The usual pat answer to all these questions
is: The firm should maximize the value of current shareholders' equity. There is a
vast literature concerning why this might be so (and what it means), which comes
down to: This makes sense (once again) only if the the firm is a price taker, where
being a price taker in this context involves many more conditions."

E-mail addresses: dierker @ihs.ac. at (shared by both authors)

Journal of the European Economic Association May 2006 4(2- 3):436-445

© 2006 by the European Economic Association

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
Dierker and Dierker General Equilibrium with Imperfect Competition 437

What should be the objective of a firm that is not a price taker? Gabszewicz
and Vial (1972) present a GE model in which firms act a la Cournot. In their
framework, shareholders are allowed to consume the goods produced by their
firm and its rivals. Like all other consumers, shareholders contribute to the gener-
ation of profits and they pay the same prices as all other consumers.1 Following
Gabszewicz and Vial, a Cournot- Walras (CW) game can be described as fol-
lows. Each firm j chooses a production plan yK These production plans are
allocated to the consumers according to their shares and determine, together with
the original initial endowments, consumers' intermediate endowments. The inter-
mediate endowments are traded in an exchange economy at Walrasian equilibrium
prices. The equilibrium prices are only determined up to a positive scalar. The
comparison of profits associated with different relative price systems constitutes
a serious conceptual difficulty referred to as the price normalization problem.
Gabszewicz and Vial use price functions to formulate the objective of a firm.
A price function ix takes a selection from the Walras equilibrium correspondence
and specifies a price normalization rule. More precisely, a price function assigns
to every profile (j1, y2, . . .) of production plans an absolute equilibrium price
vector n(yl, y2, . . .) £ M1 of the associated exchange economy. Each firm j
maximizes profits as given by iz(yl, y2, . . .)y7. CW-equilibria depend on the
selection of the price function in a critical way.
In Grodal (1984) the following embarrassing consequence of the price nor-
malization problem is shown.2 A profile of production plans is called p-dominated
if, for one of the firms, there exists another production plan that gives precisely
the same relative price system but higher profits. In a CW-equilibrium produc-
tion plans cannot be p-dominated. However, arbitrary production profiles can be
turned into CW-equilibria if they are not p-dominated. Thus, in a CW-equilibrium,
a production plan yj may lie in the interior of j 's production set YK The reader
is also referred to Bohm (1994).
H. Dierker and B. Grodal (1986) contains two examples. In the first example,
firms maximize profits on the basis of some price function. The price function
can be chosen such as to obtain any one of the following three cases: First, the
CW-game has an equilibrium in pure strategies. Second, the CW-game has no
equilibrium in pure strategies, but a mixed strategy equilibrium exists. Third, the
CW-game does not even have an equilibrium in mixed strategies. In the second
example, each firm has a single, utility maximizing owner. Again, no mixed
strategy equilibrium needs to exist.

1. In Denmark and various other countries pension funds constitute an essential column of the
pension system. A large part of the population holds shares through its contributions to pension
funds and mutual funds.

2. The article, which is written in Danish, is summarized in Grodal (1996).

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
438 Journal of the European Economic Association

In this paper, we present an example that has been developed together with
Birgit Grodal. The example illustrates, in the context of Cournot-Walras compe-
tition, the price normalization problem and a solution concept, called real wealth
maximization. This concept can be described as follows: A firm maximizes the
real wealth of its shareholders if the firm cannot choose another strategy that
makes its shareholders so rich that they can buy a larger bundle. Real wealth max-
imization is closely related to the impossibility of potential Pareto improvements
among the group of shareholders and independent of any price normalization.

2. Example: Cournot-Walras Competition

Birgit Grodal was determined to solve the question of how a firm should measure
the wealth of its shareholders without reference to utility functions. In order to
illustrate the approach taken by Grodal and her collaborators, we present and
analyze different objectives in the framework of one simple example.
There are two firms 7 = 1,2 and two consumption goods, denoted x and y.
Following Gabszewicz and Vial (1972), each firm is endowed with a certain
amount of a factor that can only be used as an input in that firm. Firm 1 is charac-
terized by its transformation curve X\ + Y\ = 9 and firm 2 by X2 + Y2 = 9.3
Inputs and costs are ignored because there is no factor market.
Firm 1 is owned by / = 1,2 and firm 2 by / = 3, 4. Each owner possesses
half of the firm. The owners have no initial endowment other than their share in
the factor endowment of their firm. Their utility functions are

u\(x,y) =xy2, u2(x,y) = xy, w3(x,v) = x2y, U4(x,y) =xy,

respectively. There are no other consumers.
At the production plan (X\, Y\) of firm 1 and prices px, py, the aggregate
demand of the owners of firm 1 for (jc, y) is

(PxXi + pyYi)[5/(12Px), l/(l2py)]

and that of the owners of firm 2 equals

(PxX2 + pyY2)U/(12px), 5/(12/7,)]

at (X2, Y2).
Firms choose their production plans simultaneously. The production plans
determine new, intermediate endowments, because each shareholder obtains half
of the production of a firm. The intermediate endowments are exchanged in a
Walrasian economy. Each firm foresees the associated Walrasian equilibrium price

3. We use capital letters to denote production plans and lowercase letters otherwise.

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
Dierker and Dierker General Equilibrium with Imperfect Competition 439

ratio perfectly. Markets in the exchange economy associated with the production
plans (X\, Y\), (X2, Y2) clear if and only if

_7Xi+5X2_7X!+5(9-y22)
Py'Px- 5Y1+1Y2 -5(9-X2) + 7F2' ()
provided that the aggregate production Y\ + Y2 of good y is positive.4
The mechanism just described does not define an oligopoly game because
the payoffs become specified only if prices are normalized. There are many ways
to normalize prices. We shall look at three specific price normalizations.

2.1. Cournot-Walras Game 1: Good x is the Numeraire

In our first specification, we measure profits in units of jc, that is to say, we consider
price systems of the form (1, py). Let the production plans (X\, Y\) and (X2, Y2)
be given. Replacing X2 by 9 - F2 and Y\ by 9 - X2, the profit of firm 1 becomes

45 + 7Xi -5F92 , 9x
n,(x,,r2) = x, + 45_5x; + 7i,M9-xj). , 9x (2)
Similarly, the profit of firm 2 equals

, 45 + 7Xi-5F?2
™'-ri>-'-ri , + 43-,xi + mri- (3)
Measuring profits in units of x can lead to an extreme underprovision of
good y. Assume that each firm believes that its opponent produces only good x.
Consider firm l's response to (X2, Y2) = (9,0). Because rii(Xi,0) = 9 +
12Xi/5, firm 1 has an incentive to increase X\ and to decrease its supply of y
as long as X\ < 3, that is, as long as the supply of y is positive. Inserting the
market-clearing price system and the corresponding demand of the owners of
firm 1 into their utility functions we obtain

_ (5 + 4Xi)3(^-9)2 _9(5 + 4Z1)2(9-X2)

Ul~ _ 10(15 + 7X02 ' U2~ 80(15 + 7Xi)
Given that firm j believes that its opponent sets its production of y equal to 0, firm
j increases its profit by producing ever smaller amounts of good y . It is apparent
from (1) that py/px tends to infinity if Y\ + Y2 tends to 0. Thus, the budget lines
of a shareholder of firm j , which pass through half of j 's output bundle, become

4. In the present example, we choose Xi and F2 as the strategic variables of the firms. Therefore,
we will often express prices, profits, and other relevant concepts as functions of (Xi , Y2).

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
440 Journal of the European Economic Association

arbitrarily steep if the supply of y approaches 0. In the limit, the budget sets of j 's
shareholders coincide with the x axis of M^_ and contain no utility maximizing
consumption bundle. On a formal level, the production plans (X\,Y\) - (3, 0)
and (X2, Y2) = (9, 0) do not constitute a Cournot-Walras equilibrium, because no
Walrasian exchange equilibrium exists. Profit maximization entails the minimum
wealth problem which causes nonexistence.5
The game possesses the (unique) CW-equilibrium X* « 2.23, F2* ^ 0.77,
in which the overproduction of the numeraire occurs in a less extreme form. We
consider this equilibrium and ask the question: Do the shareholders of a firm
want to instruct its management to change their firm's objective unilaterally? The
answer is yes. Given the strategy of its opponent, a reduction in the production
of the numeraire enables any one firm to make both of its owners better off.

2.2. Cournot-Walras Game 2: px + py - 1

We abandon the asymmetry of the price normalization but retain the assumption
that both firms measure profits in the same way. If profits are measured in units
of the bundle (jc, y) = (1, 1), then (px, py)(l, 1) = px + py = 1.
Using the market clearing condition (1), we obtain the following profit
functions in the (1,1) normalization:

Z1(45 + 7y2-5Z^) + (9-X?)(45 + 7X1-5y22)

Dl(Zl' Y2) ~ 90 + 7<X1 + r2)-5(X? + K?)
and

(9 - y|)(45 + 1Y2 - 5X\) + F2(45 + 7Xi - 5YJ)

n2(Xl' Y2) ~ 90 + 7(Xl + Y2)-5W + Yi) •
The strategy profile X* = F| ^ 115 constitutes a CW-equilibrium. Every
consumer benefits in equilibrium if the normalization px = 1 is replaced by
Px + Py = 1.
However, in spite of the symmetry of the price normalization, the equilibrium
appears unsatisfactory for the following reason. Consider, for instance, firm 1 and
assume that F| is given. If shareholder 2 alone would be in control of the firm,
he would choose X* as his optimal response to Y^ because his utility function is
symmetric with respect to x and y . Shareholder 1 , however, reaches his utility peak
at X\ ^ 0.9 < X\. The price normalization px + py - 1 acts as if shareholder
2 ruled as a dictator in firm 1. Shall the firm replace the basket (1, 1) by another
bundle? We would like to know whether the shareholders of firm 1 can be made
"richer" given the strategy of the opponent.

5. The reader is referred to Debreu (1959), Section 4.8, or to Mas-Colell, Whinston, and Green
(1995), in particular, Proposition 16.D.2.

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
Dierker and Dierker General Equilibrium with Imperfect Competition 441

We interpret this question in two different ways. The first interpretation

involves utility comparisons: Can the firm find another production plan such
that the loser can be compensated out of the winner's gain? We are searching for
an outcome which is immune with respect to potential Pareto improvements. If
no potential Pareto improvement exists no transfers need to be carried through
and the fact that transfers remain potential can cause no harm.
The second interpretation is free of any utility considerations: Is the firm able
to find another production plan such that its shareholders become rich enough to
maintain their aggregate consumption without spending all their wealth?
To answer the first question, suppose firm 1 respects the desire of shareholder
1 to reduce its output of good x a little bit. Then shareholder l's gain exceeds
shareholder 2's loss for the following reason. Because shareholder 2 reaches his
bliss point at the original strategy X* ^ 1.15 the first-order effect of a wealth
change on his utility is zero. However, shareholder 1 obtains a positive first-order
utility gain if X\ is slightly lowered.
Second, we leave all utility considerations aside and ask ourselves how an
infinitesimal change AXi < 0 of firm 1 's equilibrium quantity affects the wealth
of its shareholders. Clearly, profits FIi in the px + py = I normalization are max-
imized if firm 1 chooses its optimal response X* to y2*, hence 3i 111 (X*, F2*) - 0-
However, AXi < 0 has a negative first-order effect on the value of the aggregate
demand of firm l's shareholders. Therefore, the difference between their wealth
and their expenditures becomes positive if AXi is adopted. That is to say, the
shareholders become richer in real terms.
The above discussion leads us to the following conclusion. Even in a sym-
metric example, profits need not be measured in units of a symmetric bundle such
as (1, 1). A firm that is interested in the aggregate real wealth of its shareholders
takes the aggregate demand of its shareholders into account. The measurement
of profits becomes firm-specific. This should not come as a surprise: If a firm has
a single owner /, then the firm will maximize /'s utility.

2.3. Cournot-Walras Game 3: Real Wealth Maximization

In an economy with a functioning stock market, firms cannot control how profits
are distributed among consumers because each individual agent decides how many
shares he acquires. Accordingly, we assume that firms disregard all aspects related
to distributional justice. We assume that firms are only interested in shareholders'
aggregate real wealth, that is to say, wealth in terms of their aggregate consump-
tion. We formulate the objective of firm 1 in the framework of our example where
firms' strategies are X\ and Y2 • The analogous definition for firm 2 is obvious.

Definition 1. Strategy Xi of firm 1 maximizes the real wealth of firm l's

shareholders given the strategy Y2 of its opponent iff firm 1 possesses no other

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
442 Journal of the European Economic Association

strategy X\ such that the aggregate demand D\(X\,Y2) of firm l's shareholders
satisfies

(px,py)Di(Xl,Y2)<hi, (4)
where (px , py) and ft i denote the market clearing price system and the associated
profit at (Xi, Y2).

Real wealth maximization is closely related to the following condition: The

firm chooses its strategy such that there is no alternative strategy that makes all
its shareholders better off after a suitable redistribution among them (no potential
Pareto domination). Observe, though, that the informational requirements for
real wealth maximization are substantially less demanding than those for the
absence of potential Pareto improvements. In particular, real wealth maximization
is defined without reference to utility functions.
Inequality (4) does not depend on how prices and profits are normalized,
because the simultaneous multiplication of prices and profits by some positive
number leaves (4) unaffected. Thus, the above definition overcomes the price
normalization problem. Inequality (4) can be stated in any normalization. In par-
ticular, the normalization used to express (4) is not related to the bundle with
respect to which profits are maximized.
The above definition can easily be adapted to other frameworks. However,
the following remark must be taken into account. The present setting is quite
special since all expenditures are paid out of profits. If this framework is left
then the profit FI i in (4) must be replaced by the total wealth of the shareholders
of firm 1. The difference between the total wealth and the part derived from
profits (e.g., as generated by labor income) will, in general, depend on the firm's
action.
The first-order condition associated with real wealth maximization can be
expressed as follows: Choose any price normalization and let (px(X\,Y2),
py(X\, Y2)) and Ily(Xi, Y2) denote the market clearing prices and the profit
of firm j in this normalization, respectively. Dj(X\, Y2) denotes the aggregate
demand of firm j 's shareholders and dj the partial derivative with respect to the
j th variable. The first-order condition for real wealth maximization by both firms
can be stated as

(djPx(Xu Y2), djPy(Xi, Y2)) • Dj(Xu Y2) = 97-n,-(Xi, Y2), j = 1, 2. (5)

In the case of our example, the system (5) has the unique solution X* = YJ ^
0.997. This solution constitutes a symmetric CW-equilibrium with real wealth-
maximizing firms.
The baskets in which profits are maximized in a real wealth equilibrium are
determined endogenously. In the particular case of our example, each firm's basket
corresponds to the aggregate expenditure shares of its shareholders, which equal

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
Dierker and Dierker General Equilibrium with Imperfect Competition 443

Table 1. Equilibrium utility profiles.

Gamel 5.32 3.57 8.59 2.86

Game 2 12.74 4.87 12.74 4.87 4.41
Game 3 13.51 5.07 13.51 5.07 4.45

Walras 14.66 5.35 14.66 5.35

(5/12, 7/12) in the case of firm 1 and (7/12, 5/12) in the case of firm 2. Our
model is symmetric and p* - p* Therefore, shareholders' expenditure shares
are proportional to shareholders' consumption.

2.4. Comparison of the Three Games and Concluding Remarks

We list the equilibrium utility profiles of the three games analyzed herein and of
the perfectly competitive Walrasian equilibrium. We also list the profits in those
three case in which equilibrium prices equal /?* = p* = 0.5.
It is apparent from Table 1 that all four equilibria are Pareto comparable. In
games 2-4, equilibrium prices always equal 0.5. Profits increase if one moves
from game 2 towards game 4.6
In equilibrium, game 3 has several closely related properties that the other
games do not possess. First, each firm maximizes profits measured in units of the
aggregate demand of its shareholders. Second, if an individual firm chooses any
other strategy, the change in profits (measured with respect to any basket) does
not enable its shareholders to maintain their aggregate consumption, that is to say,
shareholders' real wealth decreases. Third, if a firm chooses another strategy, the
change in profits does not enable its shareholders to maintain their equilibrium
utility levels even if redistribution among them is allowed.
Maximization of shareholders' real wealth is related to the maximization of
shareholders' social surplus because it encompasses profits, namely, producer
surplus, as well as shareholders' consumer surplus as represented by their expen-
ditures. Producer surplus alone suffers from the price normalization problem,
but social surplus does not. Let (X*, Y£) be a CW-equilibrium with real wealth
maximization and D* the aggregate equilibrium demand of firm y's sharehold-
ers. Take any positive bundle (jq, X2) and normalize prices and profits such that
x\px + X2Py = 1. Define the social surplus of, say, firm l's shareholders with
respect to (X*, F2*) as Si(Xi) = (rii(Xi, F2*) - (px, py)D\? Then S\(X\) is

6. Remember that costs are normalized to zero. Therefore, it is not astonishing that revenues in a
perfectly competitive equilibrium exceed those in the Cournot- Walras games.
7. Because real wealth maximization is not utility based, consumers' expenditures are computed
with respect to shareholders' demand rather than shareholders' utility levels.

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
444 Journal of the European Economic Association

Figure 1. Profit minus expenditures on D* as a function of Xi .

non-positive and takes its maximum value 0 at X\ = X*. Figure 1 illustrates

this fact in the case of game 3. If we would have taken the CW-equilibrium of
game 1 or 2 instead, in which shareholders' real wealth is not maximized the
corresponding surplus function would intersect the horizontal axis at X\ = X*
and, therefore, take its maximum at some strategy X\ ^ X*.
For a more general treatment of real wealth maximization including existence
results, the reader is referred to E. Dierker and B. Grodal (1998, 1999). The rela-
tionship between real wealth maximization, surplus maximization, and potential
Pareto improvements is explored in Dierker, Dierker, and Grodal (2001).

References

Bohm, Volker (1994). "The Foundation of the Theory of Monopolistic Competition Revisited."
Journal of Economic Theory, 63, 208-218.
Debreu, Gerard (1959). Theory of Value. John Wiley.
Dierker, Egbert, Hildegard Dierker, and Birgit Grodal (2001). "Objectives of an Imperfectly
Competitive Firm: A Surplus Approach." In Economic Essays. A Festschrift for Werner
Hildenbrand, edited by Gerard Debreu, Wilhelm Neuefeind, and Walter Trockel. Springer-
Verlag.
Dierker, Egbert, and Birgit Grodal (1998). "Modelling Policy Issues in a World of Imperfect
Competition" Scandinavian Journal of Economics, 100, 153-179.
Dierker, Egbert, and Birgit Grodal (1999). "The Price Normalization Problem in Imperfect
Competition and the Objective of the Firm." Economic Theory, 14, 257-284.
Dierker, Hildegard, and Birgit Grodal (1986). "Non-Existence of Cournot-Walras Equilibrium
in a General Equilibrium Model with Two Oligopolists." In Contributions to Mathematical
Economics, edited by Werner Hildenbrand and Andreu Mas-Colell. North-Holland.
Gabszewicz, Jean, and Jean-Philippe Vial (1972). "Oligopoly 'a la Cournot' in a General
Equilibrium Analysis." Journal of Economic Theory, 4, 381-400.

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms
Dierker and Dierker General Equilibrium with Imperfect Competition 445

Grodal, Birgit (1984). "Profit Maximizing Behavior in General Equilibrium Models with
Imperfect Competition" [in Danish]. Economic Essays, 28, 79-89.
Grodal, Birgit (1996). "Profit Maximization and Imperfect Competition." In Economics in a
Changing World 2, edited by Beth Allen. Macmillan.
Kreps, David (1990). A Course in Microeconomic Theory. Harvester Wheatsheaf.
Mas-Colell, Andreu, Michael Whinston, and Jerry Green (1995). Microeconomic Theory.
Oxford University Press.

This content downloaded from 155.69.24.171 on Thu, 12 May 2016 14:38:35 UTC
All use subject to http://about.jstor.org/terms