An Experimental Study of Competitive Market Behavior

TL;DR: In this paper, a series of experimental games designed to study some of the hypotheses of neoclassical competitive market theory are presented. But they are intended as simulations of certain key features of the organized markets and of competitive markets generally, rather than as direct, exhaustive simulations of any particular organized exchange.

Abstract: INTRODUCTION RECENT years have witnessed a growing interest in experimental L. games such as management decision- making games and games designed to simulate oligopolistic market phenomena. This article reports on a series of experimental games designed to study some of the hypotheses of neoclassical competitive market theory. Since the organized stock, bond, and commodity exchanges would seem to have the best chance of fulfilling the conditions of an operational theory of supply and demand, most of these experiments have been designed to simulate, on a modest scale, the multilateral auction-trading process characteristic of these organized markets. I would emphasize, however, that they are intended as simulations of certain key features of the organized markets and of competitive markets generally, rather than as direct, exhaustive simulations of any particular organized exchange. The experimental conditions of supply and demand in force in these markets are modeled closely upon the supply and demand curves generated by the limit price orders in the hands of stock and commodity market brokers at the opening of a trading day in any one stock or commodity, though I would consider them to be good general models of received short-run supply and demand theory. A similar experimental supply and demand model was first used by E. H. Chamberlin in an interesting set of experiments that pre-date contemporary interest in experimental games.

Summary

II. EXPERIMENTAL PROCEDURE

• The experiments discussed in Parts III and IV have followed the same general design pattern.
• At each hypothetical price the corresponding quantity represents the maximum amount that could be sold at that price.
• Since the experiments were conducted within a class period, the number of trading periods was not uniform among of a trading period, this procedure is continued until bids and offers are no longer leading to contracts.
• Similarly, each seller, the authors may imagine, has "overnight" acquired a fresh unit of the commodity which he desires to sell in period 2 under the same minimum price conditions as prevailed in period 1.
• Furthermore, the only way that a real marketer can obtain knowledge of market conditions is to 6.

OF EXPERIMENTAL RES-ULTS

• The supply and demand schedule for each experiment is shown in the diagram on the left of Charts 1-10.
• From the results it is seen that the earlier conjecture concerning the effect of a divergence between buyer and seller rent on the approach to equilibrium is confirmed.
• The average contract price in trading period 4 is $10.90, only $0.15 above the predicted equilibrium.
• The competition of sellers pushed the offer prices lower and the remaining buyers made contracts at prices ($2.35, $2.00, and $2.00) near or below the equilibrium price.
• My first empirical investigation is concerned with the measuremet of the equilibrating tendencies in these markets and the ability of supply and demand theory to predict the equilibrium price in each experiment.

V. RATIONALIZATION OF THE EXCESS-RENT HYPOTHESIS

• H-aving provided a tentative empirical verification of the hypothesis that price in a competitve market tends to rise or fall in proportion to the excess buyer plus seller rent corresponding to any contract price, it remains to provide some theoretical rationale for such a hypothesis.
• Such resistance may tend to give way, even where the rents on one side or the other are very small, if it becomes clear that such rents are unattainable.
• A particularly interesting aspect of the excess-rent hypothesis is that it leads naturally to an interesting optimality interpretation of the static competitive market equilibrium.
• It is artificial in the sense that after an Invisible Hand has led us to its maximization, the authors need not necessarily attach any social welfare significance to the result."".
• Similarly, the market mechanism provides a set of "laws" which minimizes the "wasteful" payment of excessive economic rent.

VI. SUMMARY

• It would be premature to assert any broad generalizations based upon the ten experiments the authors have discussed.
• Yet conclusions are important for purposes of specifying the exact character of any findings, whether those findings are ultimately verified or not.
• The following tentative conclusions are offered concerning these experiments: 1. Even where numbers are "small," there are strong tendencies for a supply and demand competitive equilibrium to be attained as long as one is able to prohibit collusion and to maintain absolute publicity of all bids, offers, and transactions.
• Changes in the conditions of supply or demand cause changes in the volume of transactions per period and the general level of contract prices.
• Some slight evidence has been pro-vided to suggest that a prediction of the static equilibrium of a competitive market requires knowledge of the shapes of the supply and demand schedules as well as the intersection of such schedules.

III. A PILOT EXPERIMENT IN "SHORT-RUN ) AND " LONG-RUN " EQUILIBRIUM

• An important characteristic of the tenl experiments discussed in this paper was the absence of any quantity-adjusting decision-making behavior on the part of either buyers or sellers.
• Such experiments represent the simulation of markets for commodities which do not have to be delivered or even produced until after the sale contract is executed.
• Hence, the possibility of distress sales, leading to losses by sellers, is ruled out by experimental design.
• The results of one pilot experiment to simulate this process is shown in Chart 12.
• The significant new element in this experiment was giving all sellers the option at the beginning of each trading period of enterinig the market or remaininig "out of produc-.

Figures

TABLE 4

TABLE 1

TABLE 3

Full text

Chapman University
Chapman University Digital Commons
"-,-+("0 "2*16/1("*$0 ,#$0$ /"' "-,-+("0

An Experimental Study of Competitive Market
Behavior
Vernon L. Smith
Chapman University30+(1'"' .+ ,$#2
-**-41'(0 ,# ##(1(-, *4-/)0 1 ';.#(&(1 *"-++-,0"' .+ ,$#2$"-,-+("0 /1("*$0
 /1-%1'$ "-,-+(":$-/6-++-,0
:(0/1("*$(0!/-2&'11-6-2%-/%/$$ ,#-.$, ""$00!61'$"-,-+("0 1' .+ ,,(3$/0(16(&(1 *-++-,01' 0!$$, ""$.1$#%-/(,"*20(-,
(,"-,-+("0 "2*16/1("*$0 ,#$0$ /"'!6 , 21'-/(7$# #+(,(01/ 1-/-%' .+ ,,(3$/0(16(&(1 *-++-,0-/+-/$(,%-/+ 1(-,.*$ 0$
"-,1 "1 * 2&'1(,"' .+ ,$#2
$"-++$,#$#(1 1(-,
+(1'$/,-,8,5.$/(+$,1 *12#6-%-+.$1(1(3$ /)$1$' 3(-/9 Journal of Political Economy

An Experimental Study of Competitive Market Behavior
Comments
:(0 /1("*$4 0-/(&(, **6.2!*(0'$#(, Journal of Political Economy3-*2+$(002$(,
Copyright
,(3$/0(16-%'(" &-
:(0 /1("*$(0 3 (* !*$ 1' .+ ,,(3$/0(16(&(1 *-++-,0 ';.#(&(1 *"-++-,0"' .+ ,$#2$"-,-+("0 /1("*$0

THE JOURNAL
OF
POLITICAL
ECONOMY
Volume LXX
APRIL 1962
Number 2
AN
EXPERIMENTAL
STUDY
OF
COMPETITIVE
MARKET
BEHAVIOR'
VERNON
L.
SMITH
Purdue University
I. INTRODUCTION
RECENT years
have
witnessed
a grow-
ing
interest
in
experimental
games
such as management
de-
cision-making
games and
games
designed
to simulate
oligopolistic
market
phenom-
ena.
This article
reports
on a
series of
experimental
games designed
to study
some of the
hypotheses
of
neoclassical
competitive
market
theory.
Since
the
organized
stock,
bond,
and commodity
exchanges
would
seem
to
have
the best
chance
of
fulfilling
the conditions
of
an
operational
theory
of
supply
and de-
mand,
most
of these
experiments
have
I
The experiments
on
which
this
report
is based
have been
performed
over
a six-year
period
begin-
ning in 1955.
They are
part
of a continuing
study,
in which the
next
phase is
to include
experimentation
with monetary
payoffs
and
more complicated
ex-
perimental
designs
to which passing
references
are
made
here
and there
in
the present
report.
I
wish
to thank
Mrs. Marilyn
Schweizer
for
assistance
in
typing
and
in
the
preparation
of charts
in this
paper,
R. K. Davidson
for performing
one of
the
experi-
ments
for
me,
and
G. Horwich,
J.
Hughes,
H.
Johnson,
and J.
Wolfe
for reading an earlier
version
of the paper
and enriching
me
with their comments
and encouragement.
This
work was supported
by
the
Institute
for
Quantitative
Research
at
Purdue,
the
Purdue
Research
Foundation,
and in part
by
National Science
Foundation,
Grant No.
16114,
at
Stanford University.
been designed
to simulate,
on
a
modest
scale,
the multilateral
auction-trading
process
characteristic
of these
organized
markets.
I
would
emphasize,
however,
that
they
are
intended
as
simulations
of certain
key
features
of the organized
markets
and
of competitive
markets
gen-
erally,
rather
than
as
direct,
exhaustive
simulations
of any particular
organized
exchange.
The experimental
conditions
of
supply
and demand
in force
in
these
markets
are modeled
closely
upon
the
supply
and demand
curves
generated
by
the
limit
price
orders
in the hands
of
stock
and
commodity
market brokers
at the
opening
of a
trading
day
in
any
one
stock
or
commodity,
though
I
would
consider
them to be
good
general
models
of received short-run
supply
and demand
theory.
A
similar
experimental
supply
and
demand
model
was
first used
by
E.
H.
Chamberlin
in
an interesting
set
of
experiments
that
pre-date
contem-
porary
interest
in
experimental
games.2
2
"An
Experimental
Imperfect
Market,"
Journal
of
Political
Economy,
LVI
(April,
1948),
95-108.
For an
experimental
study
of bilateral
monopoly,
see
S.
Siegel
and L. Fouraker,
Bargaining
and
GJoitp
Decision
SMaking
(New
York: McGraw-Hill
Book
Co.,
1960).
ll
This content downloaded from 206.211.139.204 on Thu, 2 Oct 2014 19:45:24 PM
All use subject to JSTOR Terms and Conditions

112
VEl"RN(N L. SMI'Tll
Chamberlin's paper
was highly sugges-
tive
in
demonstrating the
potentialities
of
experimental
techniques in the
study
of
applied
market
theory.
Parts
II
and III of
this paper
are
devoted
to a descriptive
discussion of
the
experiments and some
of their
detailed
results.
Parts IV and
V present an em-
pirical
analysis of
various
equilibrating
hypotheses and a
rationalization of
the
hypothesis found to be
most successful
in these
experiments.
Part
VI provides a
brief
summary
which the
reader
may
wish
to consult
before
reading
the
main
body of the paper.
II.
EXPERIMENTAL
PROCEDURE
The
experiments
discussed in
Parts
III
and
IV have
followed the same gen-
eral
design pattern. The
group
of
subjects
is divided
at random
into two
subgroups,
a
group
of
buyers
and
a
group
of
sellers.
Each
buyer receives a
card
containing
a
number,
known
only
to
that
buyer,
which
represents
the maximum
price
he
is
willing to pay for
one unit of
the
fictitious
commodity.
It
is
explained
that
the
buyers
are not to
buy
a unit of the
commodity at a price
exceeding
that
appearing
on
their
buyer's card;
they
would be
quite
happy
to
purchase
a
unit at
any price
below
this
number-the
lower
the
better;
but, they would
be
entirely willing
to
pay just this price
for
the
commodity
rather
than have
their
wants
go
unsatisfied.
It is further ex-
plained that each
buyer should
think
of himself
as
making
a
pure profit equal
to the difference
between
his actual
con-
tract
price
and the maximum reserva-
tion
price
on
his card. These
reservation
prices
generate
a
demand curve such
as
DD
in
the
diagram
on
the
left
in
Chart
1.
At each
price
the
correspond-
ing
quantity represents the maximum
amount that could
be
purchased
at
that
price.
Thus,
in Chart
1,
the
highest
price
buyer
is willing
to
pay
as
much
as $3.25
for one
unit. At
a
price
above
$3.25
the demand
quantity
is
zero,
and
at
$3.25
it cannot
exceed
one
unit.
The
next
highest
price
buyer
is
willing
to
pay
$3.00.
Thus,
at $3.00
the demand
quantity
cannot
exceed
two
units.
The
phrase
"cannot
exceed"
rather
than
"is"
will be
seen
to be of no
small
impor-
tance.
How
much is
actually
taken
at
any price
depends
upon
such
important
things
as
how
the
market
is organized,
and
various
mechanical and
bargaining
considerations
associated
with
the
offer-
acceptance
process.
The demand curve,
therefore,
defines
the
set
(all
points
on
or
to the left of DD)
of
possible
demand
quantities
at
each, strictly
hypothetical,
ruling
price.
Each
seller
receives
a card
containing
a
number,
known only
to that seller,
which
represents
the minimum
price
at
which
he
is
willing
to
relinquish
one unit
of the
commodity.
It
is
explained
that
the
sellers
should
be willing
to sell
at
their
minimum
supply
price
rather
than
fail
to make a
sale,
but
they
make
a
pure
profit
determined
by
the
excess
of
their
contract
price
over their
mini-
mum
reservation
price.
Under
no con-
dition
should
they sell
below
this
mini-
mum.
These
minimum seller
prices
gen-
erate
a
supply
curve such as SS
in
Chart
1.
At
each
hypothetical
price
the
cor-
responding
quantity
represents
the
maxi-
mum amount
that
could be sold
at
that
price.
The
supply
curve,
therefore,
de-
fines
the set
of
possible
supply
quantities
at each
hypothetical
ruling price.
In
experiments
1-8
each
buyer
and
seller
is allowed
to
make
a
contract
for
the
exchange
of
only
a
single
unit
of
the
commodity
during
any
one
trading
or market
period.
This
rule was
for the
sake
of
simplicity
and was
relaxed
in
This content downloaded from 206.211.139.204 on Thu, 2 Oct 2014 19:45:24 PM
All use subject to JSTOR Terms and Conditions

EXPERIMENTAL
STUDY OF
COMPETITIVE MARKET
BEHAVIOR 113
subsequent
experiments.
Each
experiment was
conducted over
a
sequence
of
trading periods five to
ten minutes
long
depending upon the
number
of
participants
in
the test
group.
Since
the
experiments
were
conducted
within a class
period,
the
number of
trading periods
was not uniform
among
CHART
1
TEST
1
$4.00
P
X_.
$4.00
ta80
- 3.20
32t
to'
2.60
2A0
,20
2.40
2.20.
tao.2
too
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-1.60
too
-
1.60
140UANT I TY
TRANSACTION
NU
R
I
PERIOD
)1.40
1.20
1
.20
too10.0
oc1LS. ca=sa Oca52 0C=5.5
a1=ss
1
.60
D
.60
.40 1
- .40
.20-
PERIOD
1
PERIOD
2
PERIOD
S
P ERIOD
4
PERIOD
5
.20
ol
1
2
3 4
5
6 7 8
9 10
11
12
012 3
4
51
23
4
512 3
4
5123
4
5
671
23 4 56
QUANTITY
TRANSACTION
NUMBER
1SY
PERIOD)
the various
experiments.
In
the
typical
experiment, the
market
opens
for
trad-
ing
period 1. This
means that
any
buyer
(or
seller)
is
free
at
any
time to
raise
his
hand
and
make a
verbal
offer to
buy (or
sell)
at
any
price
which
does
not
violate his
maximum
(or
minimum)
reservation
price.
Thus,
in
Chart
1,
the
buyer
holding the
$2.50
card
might raise
his
hand
and
shout,
"Buy
at
$1.00."
The
seller
with
the
$1.50
card
might
then
shout,
"Sell
at
$3.60."
Any
seller
(or
buyer)
is
free to
accept
a
bid (or
offer),
in
which
case a
binding
contract
has been
closed, and the
buyer and
seller
making the
deal drop
out of the market
in the sense
of no longer
being
permitted
to make
bids, offers, or
contracts
for
the
remainder of
that
market period.3
As
soon
as a
bid or offer
is
accepted,
the
contract price
is
recorded together
with
the
minimum
supply price
of the
seller
and
the
maximum
demand
price
of
the
buyer
involved in the
transaction. These
observations
represent the
recorded data
of
the
experiment.4
Within the
time
limit
3
All
purchases are
for
final
consumption.
There
are
no
speculative
purchases
for
resale in
the
same
or
later
periods.
There
is
nothing,
however, to
pre-
vent
one from
designing an
experiment in
which
purchases for
resale
are
permitted if
the
objective
is to
study
the
role of
speculation in
the
equilibrating
process.
One
could, for
example, permit
the
carry-
over
of
stocks
from one
period
to
the next.
4
Owing to
limitations
of
manpower and
equip-
ment
in
experiments
1-8,
bids
and
offers
which
did
not
lead to
transactions
could
not be
recorded.
In
subsequent
experiments a
tape
recorder
was
used
for
this
purpose.
This content downloaded from 206.211.139.204 on Thu, 2 Oct 2014 19:45:24 PM
All use subject to JSTOR Terms and Conditions