Why Should the Economy be Competitive?

Abstract

Is a competitive free market the most efficient way to maximize welfare
and to equally allocate rare resources among economic agents? Economists usually
tend to think this is the case. This paper presents a preliminary attempt through an
object-oriented multi-agent model to address this question. Agents which are alternatively
producers, sellers, buyers and consumers participate in a market to increase
their welfare. Two market models are compared: the competitive market, which is
a double auction market in which agents attempt to outbid each other in order to
buy and sell products, and the random market, in which products are allocated in a
random manner. The comparison focuses on wealth creation (a common study) and
above all on inequalities in resources allocation (much less frequently addressed).
We compute and compare in both cases the money and utility generated within the
markets as well as a Gini index that characterizes the degree of inequality in the allocation
of these values among agents. In contrast with earlier works, we also compare
welfare creation and distribution with more or less intelligent agents who use more
or less information provided by the market. The results of hundreds of simulations
are analyzed to find that, as is well known, the competitive market creates more
value overall but, and generally less accepted and discussed by economists, at the
expense of a much greater inequality. We further find that whereas the use of market
information by the producers leads to less inequality in both types of markets,
the inequality induced by the competitive market depends on the behavior of the
agents, suggesting that it is both the institutions and the agents that foster inequality
in competitive market structures.

Full text

Page 1

Electronic copy available at: http://ssrn.com/abstract=1945660
Why should an economy be competitive ?
Hugues Bersini
IRIDIA - Universite Libre de Bruxelles
CP 194/6 - 50, av. Franklin Roosevelt,
Bruxelles 1050 - Belgium
Nicolas van Zeebroeck
ECARES - Solvay Brussels School of Economics and Management
Universite Libre de Bruxelles
CP 114 - 50, av. Franklin Roosevelt,
Bruxelles 1050 - Belgium
This paper provides a new look at the well-known trade-off be-
tween efficiency and equality with an agent-based model. By way of
a computer program, our study simulates the interactions of agents
producing and trading goods within different market structures and
looks at the resulting production and distribution of welfare among
agents at the end of an arbitrarily large number of iterations. Two
market mechanisms are compared: the competitive market (a double
auction market in which agents outbid each other in order to buy and
sell products) and the random market (in which products are allocated
randomly). Our results first confirm that the superior efficiency of the
competitive market comes at a very high price in terms of inequality,
compared with the random market. We further explore the effect of
agent rationality in production and auctioning (i.e. different informa-
tion sets used or not by the agents in making their choice) and observe
that although rationality affects efficiency only at the margin, inequal-
ities can be very strongly affected by the behaviour of the agents. This
latter result suggests that market mechanisms can ensure optimal ef-
ficiency under certain constraints, but that the degree of inequality
emerging from a certain market design can strongly depend on the
rationality of the agents.
1. Introduction
Economic tradition holds that competition and free markets are the
best way to create value. Adherents to Adam Smith’s exposition of
the invisible hand and famous historical advocates of competitive or
perfectly concurrent free market, such as Friedrich Von Hayek, Milton
Friedman and many others, have usually seen in this type of decen-
tralized and self-organized economical exchanges between competitive
sellers and buyers the most efficient way to maximize social welfare.
Complex Systems, volume (year) 1–1+; year Complex Systems Publications, Inc.

Page 2

Electronic copy available at: http://ssrn.com/abstract=1945660
2
H. Bersini, N. van Zeebroeck
More recently, Gode and Sunder [3] have shown that highly competitive
market structures such as double auctions lead to efficient product al-
location and price setting, even when the agents have zero-intelligence
(i.e. place offers and bids in a purely random way). They conclude
that imposing a budget constraint to the agents is sufficient to maxi-
mize the allocative efficiency of these auctions, even with irrational (or
Zero-Intelligence) agents.
Not surprisingly, a competitive system potentially likely to promote
the winners at the expense of the losers may have little chances to
equally distribute wealth. Competition in the economy is inherent to
the nature of the law of supply and demand that rules the dynamics of
these markets, calibrating the products price such as to adjust demand
to supply. As a result of this competition among agents, the price of
a rare product to acquire should increase, favoring the richest buyers,
and the price of an abundant product to sell decreases, favoring the
most performing sellers (who propose lower prices). One first classical
counterargument to this apparent source of inequality is that, by com-
peting, the sellers will finally provide products at cheaper price to the
poorest buyers, and reciprocally, the buyers, by competing, will allow
the unqualified sellers to also profit from the products they sell.
These factors translate into the well-known trade-off between effi-
ciency and equality (see e.g. [9, 7]). Although the existence of such
a trade-off has been discussed for a while, relatively less attention
has been devoted to questions such as: Which features of competi-
tive markets trigger or foster inequalities among agents? How does
the behavior of agents affect inequality? By giving up competitive
market structures, would the loss in efficiency be proportional to the
gain in equality? More generally, whereas some scholars have analyzed
inequality within competitive market structures, fewer works have in-
vestigated the efficiency and egalitarian properties of alternative mar-
ket mechanisms. Besides, most economists still advocate for matching
mechanisms that maximize efficiency and compensate for the resulting
inequality through redistributing processes, without questioning much
this philosophical choice.
A succession of theoretical models has shown that it is possible to
maximize the efficiency in product allocation through price adjust-
ments. However, although all agents benefit from the exchanges at the
aggregate level, individual welfare still depends on the agent’s tastes
distribution, budget constraints and production skills, and not much is
told about the welfare distribution among the agents. Allocative effi-
ciency and Pareto optimality have nothing to do with equality among
agents. As a matter of fact there is a long going philosophical dispute
between the utilitarians (maximizing cumulated welfare) and the egal-
itarians (reducing welfare variability) [4], since both objectives (while
equally ethically justified) appear to be antagonistic in many social
Complex Systems, volume (year) 1–1+

Page 3

Why should an economy be competitive ?
3
contexts. Clearly a competitive economy looks towards the utilitarian
side.
With regard to the egalitarian perspective, little attention has been
paid so far in the literature to the side-effects of competitive market
structures on welfare distribution among agents. This is a difficult
question to address from an analytical point of view. We believe that
conducting such a comparative analysis through a more classical math-
ematical approach would be very complicated if not impossible with
the mathematical instruments usually found in economy. In this paper,
a software stylized model is presented in order to compare two styles
(and thus purposefully unrealistic) of market structures: competitive
(a double auction market in which buyers and sellers compete by trying
to outbid each other) v. random (or distributive) (where the matching
between buyers and sellers is done in a purely random way insofar as
certain constraints are met). The two market structures are compared
along two main dimensions: the welfare produced as a result of the
agents’ interactions (in terms of utility and money), and the degree of
inequality in welfare allocation between agents.
The application of agent-based modeling to economic problems is
however not new. Agent-based Computational Economics, or ACE,
is described by Tesfatsion as “the computational study of economies
modeled as evolving systems of autonomous interacting agents” [11].
Instead of a top down, analytical approach, this type of models stud-
ies economies from the bottom up, by simulating a large number of
interacting agents and observing what kind of phenomena do emerge.
Agent-based models are becoming more and more prevalent in eco-
nomics. A few examples are illustrated in [6, 5, 12].
Many mechanisms such as the parallel interactions among simple
agents (since just motivated by profit), the reactivity of these agents
to the stigmergic effect of price, buying less when prices increase and
pushing prices to decrease by selling more of the most wanted products,
the self-organized stabilization of prices which equilibrate supplies and
demands, testify of the clear connections existing between the sort of
computational models popular in the complex systems or artificial life
literature and the working of free markets. As a matter of fact, John
Holland was among the authors of a 15 years old paper entitled ”Artifi-
cial Economic Life: A Simple Model for a Stockmarket” [8] where they
state in the introduction: ”This stockmarket model may also be seen
as a case-study in artificial life; from a random soup of simple rules an
intelligent system spontaneously arises”. As reminded by John Cassidy
[1], Friedrich Hayek, the key historical figure of free market economy,
was a sort of complex system pioneer, when stating that market prices
are primarily a means of collating and conveying information (just like
insects pheromone) that would be totally impossible to replicate for a
centralized cognitive planner.
Complex Systems, volume (year) 1–1+

Page 4

4
H. Bersini, N. van Zeebroeck
”Hayek’s metaphor of the market as a ”system of telecommunica-
tions” is more direct and specific. It helps explain how market works -
via the transmission of price signals - and why they are so difficult to
replicate. The most significant fact about this system is the economy
of knowledge with which it operates, or how little the individual partic-
ipants need to know in order to be able to take the right action. Only
the most essential information is passed on and passed on only to those
concerned”.
The paper that most closely relates to ours is Gode and Sunder
[3], who look at the dispersion of profits among agents in the human
v. zero-intelligence markets and conclude that contrarily to allocative
efficiency that can be maximized by the market structure itself, profit
dispersion may depend on the behavior of the agents. They however do
not investigate the impact of human behavior on utility dispersion. The
present article departs from Gode and Sunder’s work in different ways.
First, Gode and Sunder look at the difference in allocative efficiency of
a single market structure when two different types of agents interact,
human agents v. Zero-Intelligence machine agents who place bids and
offers randomly.In contrast, we look at the same sets of machine
agents who interact within two different types of market structures:
competitive v. random. Second, we not only look at welfare creation
(or allocative efficiency), but most importantly at welfare distribution
among agents, in terms of both utility and money. Finally, we also
investigate the impact of different agents behaviors in the different
markets on welfare creation and distribution.
In practice, a computer program will be presented in the following,
implementing simple rules for the agents to follow as well as a frame-
work which allows them to interact, while logging facilities collect data
on the transactions executed and the evolution of the different metrics
in the model. The agents are then set free in numerous simulations,
and the resulting metrics are compared across simulations. Our first
sets of results show that a competitive market structure such as a dou-
ble auction consistently leads to much higher welfare at the aggregate
level than a random matching market. But they reveal considerable
differences in the distribution of utility and money between the two
market structures, with the competitive market leading to much more
unequal distribution of welfare between agents. We then explore the
impact of informed v. zero-intelligence producers. One obvious proof
of the poor efficiency of a random production is that an important
percentage of simulations just fail since no matching occurs between
what the consumers want most and what is being offered by the pro-
ducers. We find that even in the most competitive market, ignorance
of the producers leads to slightly more utility for the consumers but
at the expense of much more money being wasted and therefore makes
both markets much less efficient. We finally investigate different be-
Complex Systems, volume (year) 1–1+

Page 5

Why should an economy be competitive ?
5
haviors of our machine buyers and sellers in selecting the products they
want to place bids or offers for. We observe that although welfare cre-
ation is generally equal, the degree of inequality in utility and money
distribution with the competitive market significantly depends on the
rationality of the agents.
In the coming section, the classes of the object-oriented model will
be described. Then the experimental results of the cumulated distri-
bution of wealth will be compared across the two types of economy:
competitive and random. As usual with these kinds of artificial models
(such as with classical game theory models and many complex sys-
tem models), the simulation presented here is not intended to depict
any precise reality but needs to be construed as a software-thought
experiment, namely the conception and the execution of virtual worlds
helping to understand in outlines the behavior of a purposefully cari-
catured reality.
2. The Model
The model implemented in C# maps elegantly to an object oriented
model with the distinct responsibilities distributed through the differ-
ent classes. These classes and their relationships can be seen in the
simplified class diagram in figure 1.
2.1 The World
The model’s different components all live within a structure called the
world. The World contains all the agents as well as the market. Each
world has one market, either a competitive double-auction market or
a random market, a series of agents and some world specific settings
such as the initial endowment of the agents, the number of different
products the agents can make and trade. A given number of products
are bought and sold. The world is not limited in the number of units
for each product, but each transaction concerns only one unit of the
product. Each world corresponds to one simulation. Worlds always
come in pairs with equal initial settings, but one taking care of the
competitive market and the other one of the random market.
2.2 The agents
The agents are the main actors of the model, they are the imaginary
people who produce, consume and trade goods driving the model’s
markets. Each agent behaves alternatively according to its integrated
Producer, Consumer, Buyer and Seller classes. So each agent plays
the four roles in turn. Agents are defined by several key numbers,
including:
Complex Systems, volume (year) 1–1+