Two-player asymmetric contests with ratio-form contest success functions.
Baik, Kyung Hwan
I examine players' equilibrium effort levels in two-player
asymmetric contests with ratio-form contest success functions. I first
characterize the Nash equilibrium of the simultaneous-move game. I show
that the equilibrium effort ratio is equal to the valuation ratio, and
that the prize dissipation ratios for the players are the same. I also
show that the prize dissipation ratio for each player is less than or
equal to the minimum of the players' probabilities of winning at
the Nash equilibrium and thus never exceeds a half. Then I examine how
the equilibrium effort ratio, the prize dissipation ratios, and the
players' equilibrium effort levels, respond when the players'
valuations for the prize or their abilities change. (JEL D72, C72)
I. INTRODUCTION
A contest is a situation in which players compete with one another
by expending irreversible effort to win a prize. Typical examples are
various types of rent-seeking contests: competition among firms to win a
monopoly rent secured under government protection or by a government
procurement contract, competition between domestic and foreign firms to
obtain governmental trade policies favorable to them, competition among
firms to acquire a rent generated by rights of ownership to an import
quota, and competition among firms to capture rents created by
governmental decisions to establish tariffs or other trade barriers.
Other examples of contests include auctions, patent races, research and
development (R&D) competition among firms, litigation, competition
for jobs among job candidates, competition among candidates to win
promotion to higher ranks, election campaigns between political
candidates, and competition between local governments to invite business
firms, government institutions, or government-owned corporations into
their districts.
Naturally, due to their prevalence and importance in economies,
such contests have been studied by many economists: Loury (1979), Lee
and Wilde (1980), Tullock (1980), Rogerson (1982), Rosen (1986),
Appelbaum and Katz (1987), Dixit (1987), Hillman and Riley (1989),
Hirshleifer (1989), Ellingsen (1991), Nitzan (1991, 1994), Krishna and
Morgan (1997), Che and Gale (1998), Hurley and Shogren (1998), and
Konrad (2000), to name a few. In this vast literature on the theory of
contests, one of the main issues is: How much effort do the players
exert in pursuit of the prize? Indeed, it is of great interest because
the players' effort levels determine the profitability of the
players' engaging in the contest and, in some cases, they are
revenues collected by the contest organizer or bribes given to
government officials. Furthermore, they account for other important
outcomes of the contest. For example, in a rent-seeking contest, efforts
expended by the players are viewed as social costs due to rent-seeking
activities, so that total effort level is a measure of economic
efficiency. In an R&D contest, effort levels expended by the
players--these are R&D expenditures of the firms--determine the
expected date of invention.
This article also addresses the issue: How much effort do the
players exert in pursuit of the prize? But it differs from previous
research by dealing with contests with ratio-form contest success
functions.1 Specifically, the novelty of this article is to consider
two-player asymmetric contests in which each player's probability
of winning is a function of the ratio of the two players' effort
levels.
Two-player contests with ratio-form contest success functions or
two-player contests that can be best modeled with ratio-form contest
success functions are easily observed in the real world. Examples
include various types of two-player rent-seeking contests, litigation
between a plaintiff and a defendant, election campaigns between two
parties or candidates, and R&D competition between two firms.
Consider, for example, a rent-seeking contest in which two firms,
potential monopolists, compete with each other to win a government
monopoly franchise contract. The firms expend outlays to influence the
government officials who have the authority to grant the monopoly
franchise. I believe this contest can be best modeled with ratio-form
contest success functions--that is, it is natural that each firm's
probability of winning the monopoly franchise should be a function of
the ratio of the two firms' "investments."
Another example is a contest in which a local government has a
budget for building a bridge and two communities compete against each
other to win that budget. (2) The local government uses a lottery-like
winner-selecting mechanism in which two participating communities
purchase as many lottery tickets as they want from the local government,
and the winner is selected by drawing one ticket out of the tickets
sold. The collected money goes to the general fund of government
revenues. In this contest, the probability that a community wins the
budget is equal to the number of tickets the community holds divided by
the total number of tickets sold, which means that the communities face
the simplest form of ratio-form contest success functions. (3)
This article focuses on the effects of asymmetries between the two
players on the equilibrium effort ratio, the prize dissipation ratios,
and the players' equilibrium effort levels. The asymmetries between
the players arise because of their different valuations for the prize or
their different abilities to convert effort into probability of winning
or both. For example, in the rent-seeking contest, the firms may have
different valuations for the monopoly franchise, and their abilities to
influence the government officials may differ. Especially, imagine the
case in which the government periodically reassigns the monopoly
franchise. In a given period, the previous franchise holder may value
the franchise more highly and may have more abilities to influence the
government officials than its rival, because it has invested resources
and has experience in that particular industry and because it has been
able to establish a relationship with the government officials. In the
budget-seeking contest, the communities may have different valuations
for winning the budget: The benefit of one community from building its
own bridge may be larger than that of the other. On the other hand, the
local government may use a biased winner-selecting mechanism by charging
the communities different prices for the lottery ticket.
This article considers the case where the two players choose their
effort levels simultaneously and independently. I first find and
characterize the Nash equilibrium of this simultaneous-move game. I show
that at the Nash equilibrium, the ratio of the players' effort
levels (of player 1 to player 2)--the equilibrium effort ratio is equal
to the corresponding valuation ratio, and that the prize dissipation
ratio for one player the proportion of his or her equilibrium effort
level to his or her valuation for the prize is equal to that for the
other. These results imply that a high-valuation player exerts more
effort than a low-valuation player, and that a high-valuation winner--a
high-valuation player as the actual winner--earns a greater net payoff
than a low-valuation winner. I also show that the prize dissipation
ratio for each player is less than or equal to the minimum of the
players' probabilities of winning at the Nash equilibrium and thus
never exceeds a half: Each player's equilibrium effort level never
exceeds half his or her valuation for the prize. This means that the
equilibrium total effort level never exceeds the valuation of a
high-valuation player or, more tightly, the average of the players'
valuations. Based on this, I may argue that over-dissipation of the
prize never occurs in this contest.
Then I examine how the equilibrium effort ratio, the prize
dissipation ratios, and the players' equilibrium effort levels
respond when the players' valuations for the prize or their
abilities change. Examining the effects of changing the players'
valuations for the prize, I show the following. If the players'
valuations for the prize increase proportionately, then their
equilibrium effort levels increase proportionately, and furthermore they
increase at the same rate as the players' valuations. Therefore, in
this case the equilibrium effort ratio and the prize dissipation ratios
remain unchanged. Next, if the valuation ratio changes, then the
equilibrium effort ratio changes in the same direction as the valuation
ratio does, but the prize dissipation ratios change in either the same
or the opposite direction. Finally, I show that when just one
player's valuation for the prize changes, his or her equilibrium
effort level changes in the same direction as his or her valuation does.
Examining the effects of changing the players' abilities on
the equilibrium effort ratio, the prize dissipation ratios, and the
players' equilibrium effort levels, I show the following. The
equilibrium effort ratio depends solely on the players' valuations
for the prize--it does not depend on the players' abilities. The
prize dissipation ratio for each player, each player's equilibrium
effort level, and the equilibrium total effort level are maximized in
the contest in which both players have equal "composite"
strength.
Baik (1994) and Nti (1997, 1999) study asymmetric contests. Balk (1994) examines the effects of asymmetries between two players on
individual and total effort levels in a model with general logit-form
contest success functions. As in this article, the asymmetries between
the players arise because of their different valuations for the prize or
their different abilities or both. Nti (1997) performs comparative
statics with respect to the number of contestants, the discount rate,
and the value of the prize, in a model with logit-form contest success
functions--more specifically, in the gametheoretic R&D model
developed by Loury (1979). He looks at their effects on individual
expenditures and individual profits, and on total expenditures and total
profits. Nti (1999) examines the effects of asymmetric valuations on
individual effort levels and individual payoffs in a model with
logit-form contest success functions that were first used in Tullock
(1980).
The article proceeds as follows. Section II develops the model and
sets up the simultaneous-move game. In Section III, I find and
characterize the Nash equilibrium of the game. In Section IV, I examine
how the equilibrium effort ratio, the prize dissipation ratios, and the
players' equilibrium effort levels respond when the players'
valuations for the prize or their abilities change. Finally, Section V
presents concluding remarks.
II. THE MODEL
Consider a contest in which two risk-neutral players, 1 and 2,
compete with each other by expending irreversible effort to win a prize.
Let [x.sub.1] and [x.sub.2] represent the effort levels expended by
players 1 and 2, respectively. Effort levels are nonnegative and are
measured in units commensurate with the prize. Let [p.sub.i] denote the
probability that player i wins the prize. I assume that each
player's probability of winning is a function of the ratio of the
two players' effort levels. More specifically, I assume the
following ratio-form contest success functions for the players:
[p.sub.1] = f([theta]) and [p.sub.2] = 1 - f([theta]), where the
function f has the properties specified in Assumption 1 below and
[theta] = [x.sub.1] /[x.sub.2]. (4)
ASSUMPTION 1. I assume that 0 <f([theta]) < 1 for all [theta]
in [R.sub.+], where [R.sub.+] denotes the set of all positive real
numbers. I also assume that f'([theta]) > 0, f"([theta])
< 0, and 2f'([theta]) + 0 f"([theta]) > 0, for all
[theta] in [R.sub.+], where f' and f" denote the first and
second derivatives of the function f.
The first part of Assumption 1 simply says that for any pair of the
players' effort levels, one for each player, each player's
probability of winning is between 0 and 1. This implies that when both
players expend positive effort levels, the player who expends a larger
effort level does not win the prize with certainty--that is, he or she
may lose the prize. (5) The first assumption of the second part--the
assumption that f'([theta]) > 0--says that each player's
probability of winning is increasing in his or her own effort level and
is decreasing in his or her opponent's effort level: In terms of
the symbols, [delta][p.sub.i]/[delta][x.sub.i]> 0 and
[delta][p.sub.i]/[delta][x.sub.j]< 0 for all [x.sub.i] >0 and for
all [x.sub.2]>0. The other two assumptions of the second part concern
the marginal effect of increasing a player's effort level on his or
her own probability of winning. The assumption that f"([theta])<
0 says that the marginal effect of player 1's effort level on his
or her probability of winning decreases as his or her effort level
increases: [delta][p.sub.1]/[delta][x.sup.2.sub.1]<0 for all
[x.sub.1] >0 and for all [x.sub.2]>0, The last assumption--the
assumption that 2f'([theta]) + [theta]f"([theta]) > 0--says
indirectly that the marginal effect of player 2's effort level on
his or her probability of winning decreases as his or her effort level
increases: [[delta].sup.2][p.sub.2]/ [delta][x.sup.2.sub.2] <0 for
all [x.sub.1] >0 and for all [x.sub.]2>0. In short, the second
part of Assumption 1 assumes that given the opponent's effort
level, each player's probability of winning is increasing in his or
her own effort level at a decreasing rate. (6) As it will be clear in
Section III, this ensures that the second-order conditions for
maximizing the players' expected payoffs are satisfied.
One example of the contest success functions which satisfy
Assumption 1 is [p.sub.1] = [sigma][x.sub.1]/ ([sigma][x.sub.1] +
[x.sub.2]) and [p.sub.2] = [x.sub.2]/ ([sigma][x.sub.1] + [x.sub.2]),
where [sigma] is a positive constant. They are extensively used in the
contest literature. (7) Another example is [p.sub.1] =
[[alpha].sub.1][x.sup.r.sub.1]/ ([[alpha].sub.1][x.sup.r.sub.1] +
[[alpha].sub.1][x.sup.r.sub.2]) and [p.sub.2] =
[[alpha].sub.2][x.sup.r.sub.2]/ ([[alpha].sub.1][x.sup.r.sub.1] +
[[alpha].sub.2][x.sup.r.sub.2]), where [[alpha].sub.1] > 0 and
[[alpha].sub.2] > 0, which are used in Tullock (1980), Hirshleifer
(1989), Skaperdas (1996), Clark and Riis (1998), and Nti (1999). The
pair of contest success functions used in Kooreman and Schoonbeek
(1997), [p.sub.1] = [(a[x.sub.1]).sup.r]/{[(a[x.sub.1]).sup.r] +
[x.sup.r.sub.2]} and [p.sub.2] = [x.sup.r.sub.2]/{[(a[x.sub.1]).sup.r] +
[x.sup.r.sub.2]}, where a>0, provides another example. (8)
The players' valuations for the prize may differ. Player 1
values the prize at [v.sub.1] and player 2 values it at [v.sub.2]. Each
player's valuation for the prize is positive and publicly known.
Let [[pi].sub.i] represent the expected payoff for player i. Then the
payoff function for player 1 is
[[pi].sub.1] = [v.sub.1]f([theta]) - [x.sub.1],
and that for player 2 is
[[pi].sub.2] = [v.sub.2](1 - f([theta])) - [x.sub.2].
I assume that the players choose their effort levels simultaneously
and independently. I also assume that all of the above is common
knowledge between the players. I employ Nash equilibrium as the solution
concept.
III. NASH EQUILIBRIUM
To obtain a Nash equilibrium of the game, I begin by considering
the maximization problems: Maximize player i's expected payoff over
his or her effort level, given player j's positive effort level.
The first-order condition for maximizing [[pi].sub.1] (without the
nonnegativity constraint) is
(1) [delta][pi].sub.1]/[delta][pi].sub.1] =
[v.sub.1]f'([x.sub.1]/ [x.sub.2])/[x.sub.2] - 1 = 0,
and that for maximizing [[pi].sub.2] is
(2) [delta][pi].sub.2]/[delta][pi].sub.2] =
[v.sub.2]f'([x.sub.1]/[x.sub.2])/[x.sup.2.sub.2] - 1 = 0,
Due to Assumption 1, player i's payoff function is strictly
concave in his or her effort level, and thus the second-order condition
for maximizing [[pi].sub.i] is satisfied.
Player i's reaction function shows his or her best response to
every possible effort level that player j might choose. Given player
j's effort level, player i's best response is defined as the
effort level that maximizes his or her expected payoff. (9) Thus player
1's reaction function is derived from condition (1). It is then
implicitly defined by
(3) [v.sub.1]f([x.sub.1]/[x.sub.2]) = [x.sub.2].
The implicit form of player 2's reaction function is
(4) [v.sub.2.][x.sub.1]f([x.sub.1]/[x.sub.2]) = [x.sup.2.sup.2]
which comes from condition (2).
A Nash equilibrium is a pair of effort levels, one for each player,
at which each player's effort level is the best response to his or
her opponent's. Thus it satisfies the players' reaction
functions--equations (3) and (4)--simultaneously. (10) Let ([x*.sub.1],
[x*.sub.2]) denote the interior Nash equilibrium. Then we have
(5) [v.sub.1] f'([x*.sub.1]/[x*.sub.2]) = [x*.sub.2]
and
(6) [v.sub.2][x*.sub.1]f'([x*.sub.1]/[x*.sub.2]) =
[([x*.sub.2]).sup.2], which yield
(7) [x*.sub.1]/[x*.sub.2] = [v.sub.1]/[v.sub.2]
or, equivalently,
(8) [x*.sub.1]/[v.sub.1] = [[x*.sub.2]/[v.sub.2].
Expression (7) says that at the Nash equilibrium, the ratio of the
players' effort levels (of player 1 to player 2) is equal to the
ratio of the players' valuations (of player 1 to player 2). In
expression (8), [x*.sub.i]/[v.sub.i] is the proportion of player
i's equilibrium effort level--which is "dissipated" in
pursuit of the prize--to his or her valuation for the prize. I call it
the prize dissipation ratio for player i. From expressions (5), (6),
(7), and (8), I obtain Proposition 1.
PROPOSITION 1. (a) The equilibrium effort ratio is equal to the
valuation ratio. (b) The prize dissipation ratios for the players are
the same. (11) (c) The equilibrium effort ratio does not depend on the
function f (d) The prize dissipation ratio for each player does depend
on the function f.
Parts (a) and (b) come mainly from two things. One is the
assumption that the two players have the ratio-form contest success
functions. The other is the solution concept of Nash equilibrium.
Because each player's probability of winning is a function of the
ratio of the two players' effort levels, the effort ratio should be
equal to the valuation ratio to satisfy the mutual-best-responses
property of Nash equilibrium. Surprisingly, the equilibrium effort ratio
is equal to the valuation ratio--equivalently, the prize dissipation
ratios for the players are the same--for any specific form of the
function f.
Parts (a) and (b) each say that player i expends
[v.sub.i]/[v.sub.j] times as much effort as his or her opponent, player
j. They also say that the proportion of player i's equilibrium
effort level to his or her valuation for the prize is equal to the
proportion of player j's effort level to his or her valuation. Each
of these implies that the players choose the same effort level if they
have the same valuation for the prize; a high-valuation player exerts
more effort than a low-valuation player. These make intuitive sense and
are consistent with what people observe in real-world contests. Indeed,
it is commonly observed that a "hungrier" player tries harder
than his or her opponent. Part (b) implies that a high-valuation
winner--a high-valuation player as the actual winners--earns a greater
net payoff than a low-valuation winner.
Parts (c) and (d) are clearly understood with the expressions,
[x*.sub.1] = ([v.sup.2.sub.1]/ [v.sub.2])f'([v.sub.1]/[v.sub.2])
and [x*.sub.2] = [v.sub.1]f'([v.sub.1]/[v.sub.2]), which are
obtained from expressions (5), (6), and (7). The two expressions show
that each player's equilibrium effort level depends on the
players' valuations, [v.sub.1] and [v.sub.2], and the function f.
Thus the prize dissipation ratio for each player should depend on the
function f On the other hand, the equilibrium effort ratio does not
depend on the function f--it depends solely on the players'
valuations--because both expressions have the common factor
f'([v.sub.1]/[v.sub.2]) and do not have any other factor related to
the function f.
Consider the case where the function f takes a form, f([theta]) =
[sigma][theta]/([sigma][theta] + 1) where [sigma]> 0, so that the
pair of contest success functions is [p.sub.1] =
[sigma][x.sub.1]/([sigma][x.sub.1] + [x.sub.2]) and [p.sub.2] =
[x.sub.2]/([sigma][x.sub.1] + [x.sub.2]). In the budget-seeking contest
introduced in Section I, the parameter [sigma] may reflect the
difference between the ticket prices charged to the communities:
Community 1 pays the local government [x.sub.1] to purchase
[sigma][x.sub.1] lottery tickets, whereas community 2 pays [x.sub.2] for
[x.sub.2] lottery tickets. In the rent-seeking contest mentioned in
Section I, the parameter [sigma] may indicate how effective the
firms' outlays are in influencing the government officials who have
the authority to grant the monopoly franchise. Indeed, looking at these
contest success functions, one may say that firm 1's outlay is
[sigma] times as effective as firm 2's. In general, the parameter
[sigma] represents player 1's abilities to convert effort into
probability of winning relative to player 2's. A value of [sigma]
greater than unity implies that player 1 has more abilities than player
2. In this case, if both players expend the same level of effort, player
1's probability of winning is greater than a half. A value of
[sigma] less than unity implies the opposite, and [sigma] = 1 implies
that both players have equal abilities. Such an ability parameter is
used in Tullock (1980), Rogerson (1982), Rosen (1986), Leininger (1993),
and Baik (1994, 1999).
Now that parts (a) and (b) hold for any specific form of the
function f, they hold for the specific form given above regardless of
the value of the ability parameter [sigma]. Furthermore, according to part (c), I obtain the same equilibrium effort ratio regardless of the
value of [sigma]--that is, the equilibrium effort ratio remains constant
when the players' abilities change. Note, however, that different
values of [sigma] lead to different pairs of the players'
equilibrium effort levels and thus lead to different pairs of the prize
dissipation ratios. To clarify these points,consider the rent-seeking
contest in which two firms compete for themonopoly franchise. Suppose
that one of the firms is replaced by another firm with different
lobbying abilities. In this case, the equilibrium effort ratio is equal
to the valuation ratio--equivalently, the prize dissipation ratios for
the firms are the same--both before and after the change. In addition,
the equilibrium effort ratio remains constant. But the equilibrium
effort level and the prize dissipation ratio for each firm, obtained
after the change, are different from those obtained before the change,
respectively.
Proposition 2 describes another interesting finding of this
article.
PROPOSITION 2. The prize dissipation ratio for each player is less
than or equal to the minimum of the players' probabilities of
winning at the Nash equilibrium and therefore never exceeds a half.
Proof The equilibrium expected payoff for player 1 is [[pi]*.sub.1]
= [v.sub.1]f([x*.sub.1]/ [x*.sub.2]) - [x*.sub.1], and that for player 2
is [[pi]*.sub.1] = [v.sub.2](1 - f([x*.sub.1]//[x*.sub.2])) -
[x*.sub.2].
Because each player's equilibrium expected payoff is
nonnegative, we obtain [x*.sub.1]/[v.sub.1] [less than or equal to]
f([x*.sub.1]/[x*.sub.2]) and [x*.sub.2]/[v.sub.2] [less than or greater
than to] 1 - f([x*.sub.1]/[x*.sub.2]) from the two expressions. The
former says that the prize dissipation ratio for player 1 is less than
or equal to player 1's probability of winning at the Nash
equilibrium. The latter says that the prize dissipation ratio for player
2 is less than or equal to player 2's probability of winning at the
Nash equilibrium. Because the prize dissipation ratios for the players
are the same, the common prize dissipation ratio must be less than or
equal to the minimum of their probabilities of winning at the Nash
equilibrium: [x*.sub.i]/[v.sub.i] [less than or equal to]
min{f([x*.sub.1][x*.sub.2]), 1 - f([x*.sub.1][x*.sub.2])}. This implies
that the common prize dissipation ratio never exceeds a half.
Given his or her opponent's effort level, each player chooses
his or her best response--the effort level that maximizes his or her own
expected payoff. Because each player's probability of winning is a
function of the ratio of the two players' effort levels, the state
of mutual best responses or, equivalently, the Nash equilibrium implies
that the prize dissipation ratios for the players are the same. The Nash
equilibrium also implies that each player's equilibrium expected
payoff is nonnegative. Utilizing these two facts, I obtain Proposition
2.
As in Baik (1994), I define a lopsided contest as a contest in
which the probability of winning for one of the players is greater than
one-half at the Nash equilibrium. As in Baik (1994), let the Nash winner
be the player who has a probability of winning greater than a half at
the Nash equilibrium and the Nash loser the player whose probability of
winning at the Nash equilibrium is less than a half. Proposition 2 then
says that in a lopsided contest, the prize dissipation ratio for each
player is less than or equal to the Nash loser's probability of
winning.
Proposition 2 says that each player's equilibrium effort level
never exceeds half his or her valuation for the prize. This implies that
the equilibrium total effort level never exceeds the valuation of a
high-valuation player or, more tightly, the average of the players'
valuations. In this sense, I might say that over-dissipation of the
prize never occurs in this contest. However, the equilibrium total
effort level may exceed the valuation of a low-valuation player.
In the literature on rent seeking, efforts expended on rent-seeking
activities are regarded as social costs that create economic
inefficiency. Therefore, examining the extent of rent dissipation is one
of the main issues in the literature. Proposition 2 establishes that
less than complete dissipation of the contested rent almost always
occurs in two-player asymmetric rent-seeking contests with ratio-form
contest success functions. This underdissipation-of-rents result is also
obtained in, for example, Tullock (1980), Hillman and Riley (1989), and
Balk and Lee (2001).
IV. COMPARATIVE STATICS
As shown in Section III, each player's equilibrium effort
level depends on the players' valuations for the prize, [v.sub.1]
and [v.sub.2], and the function f. It is interesting, then, to examine
how the equilibrium effort ratio, the prize dissipation ratios, and the
players' equilibrium effort levels respond when the players'
valuations change or when the function f takes a different specific
form. This is what I will cover in this section.
Valuations
I begin by examining the effects of changing the players'
valuations for the prize on the equilibrium effort ratio, the prize
dissipation ratios, and the players' equilibrium effort levels.
Proposition 3 is immediate from expression (7) and the fact that
[x*.sub.1]/[v.sub.1] = [x*.sub.2]/[v.sub.2] = ([v.sub.1]/[v.sub.2])
f'([v.sub.1]/[v.sub.12).
PROPOSITION 3. (a) Suppose that the players" valuations for
the prize change proportionately and thus the valuation ratio,
[v.sub.1]/[v.sub.2], remains unchanged. Then the equilibrium effort
ratio, [x*.sub.1]/[x*.sub.2], and the prize dissipation ratios,
[x*.sub.1]/[v.sub.1] and [x*.sub.2]/[v.sub.2], remain unchanged; each
player's equilibrium effort level changes in the same direction as
the players' valuations do. (b) If the valuation ratio changes,
then the equilibrium effort ratio changes in the same direction as the
valuation ratio does, but the prize dissipation ratios change in either
the same or the opposite direction.
Part (a) implies that if the players' valuations for the prize
increase at the same rate, then their equilibrium effort levels increase
at the same rate, and furthermore they increase at the very rate that
the players' valuations increase. Consider, for example, the
rent-seeking contest in which two firms compete for the monopoly
franchise. If the firms' valuations for the monopoly franchise
double, then their equilibrium effort levels double, so that the
equilibrium effort ratio and the prize dissipation ratios remain
constant.
Imagine the players' reaction curves drawn in the
[x.sub.1][x.sub.2] space. Their intersection is the Nash equilibrium.
Part (a) says that when the players' valuations rise or fall at the
same rate, the old and new Nash equilibria are located on the same
straight line that emanates from the origin, and the Nash equilibrium
associated with higher valuations lies farther from the origin.
Part (a) implies that the equilibrium total effort level changes in
the same direction as the players' valuations do. When the
players' valuations for the prize increase proportionately, player
i's best response to player j's positive effort level
increases, so that player i's reaction curve shifts outward. (12)
This leads to higher effort levels at the Nash equilibrium associated
with higher valuations. To put it in familiar words, when both
players' desire for the prize increases, so does the competition
between them that is, the players exert more effort to win the prize,
which is now more valuable than before.
Part (b) says that when the valuation ratio increases, the
equilibrium effort ratio increases, but the prize dissipation ratios
either increase or decrease. Geometrically speaking, the graph of the
relationship between the valuation ratio and the equilibrium effort
ratio is the 45[degrees] line in the corresponding two-dimensional
space. Figure 1 illustrates how the prize dissipation ratios as well as
the equilibrium effort ratio change as the valuation ratio increases, in
the case where f([theta]) = [theta]/(1 + [theta]). As the valuation
ratio, [v.sub.1]/[v.sub.2], increases from an arbitrarily small positive
number, the prize dissipation ratios first increase, reach the maximum,
and then decrease. The inflection point occurs at point A. The prize
dissipation ratios are maximized when both players have the same
valuation for the prize--that is, when [v.sub.1]/ [v.sub.2]= 1. As in
Baik (1994), 1 define the even contest as a contest in which both
players have the same probability of winning at the Nash
equilibrium--or, equivalently, as a contest in which the probability of
winning for each player is equal to one half at the Nash equilibrium. In
the case where f ([theta]) = [theta]/(1 + [theta]), it is
straightforward to see that the even contest occurs when both players
have the same valuation for the prize. Then it follows that the prize
dissipation ratios are maximized in the even contest, not in a lopsided
contest. This confirms the idea that, as contestants'
"strength" gets closer, competition gets fiercer. Note,
however, that the equilibrium effort ratio--represented by the
45[degrees] line--keeps increasing. It happens because, as the valuation
ratio increases, both players' equilibrium effort levels do not
always move in the same direction. Consider, for example, the case in
which player 1's valuation for the prize increases while player
2's remains constant. In this case, player 1's equilibrium
effort level is increasing in the valuation ratio, but player 2's
is increasing in the valuation ratio until [v.sub.1]/[v.sub.2] = 1, and
is decreasing thereafter.
[FIGURE 1 OMITTED]
I now look at the effects on the players' equilibrium effort
levels in the case of part (b)--where the valuation ratio changes. Using
the expressions [x*.sub.1] = ([v.sup.2.sub.1]/
[v.sub.2])f'([v.sub.1]/[v.sub.2]) and [x*.sub.2] =
[v.sub.1]f'([v.sub.1]/[v.sub.2]), it is straightforward to obtain
the following: If player 1's valuation for the prize increases
(decreases) but the valuation ratio decreases (increases), then player
2's equilibrium effort level increases (decreases). This says that
if player 2's valuation for the prize increases at a higher rate
than player 1's, then player 2's equilibrium effort level
increases. It also says that if player 2's valuation decreases at a
higher rate than player 1's, then player 2's equilibrium
effort level decreases. In other words, player 2 exerts more (less)
effort when his or her valuation relative to player 1's increases
(decreases). In both cases, the effects on player 1's equilibrium
effort level and the equilibrium total effort level are ambiguous.
Finally, when replacing player 1 in the conditional statement with
player 2, I cannot obtain symmetric results. This asymmetry arises
because each player's probability of winning is a function of the
ratio [x.sub.1]l[x.sub.2].
Another result that I obtain is as follows: If just one
player's valuation for the prize changes, his or her equilibrium
effort level changes in the same direction as his or her valuation does.
As in the preceding case, the effects on the other player's
equilibrium effort level and the equilibrium total effort level are
ambiguous. Baik (1994) conducts comparative statics with respect to the
valuation parameter in contests with logit-form contest success
functions. He shows that, as one player's valuation for the prize
increases from an arbitrarily small positive number (while the other
player's remains unchanged), the other player's effort level
first increases, reaches the maximum in the even contest, and decreases
thereafter. He also shows that the equilibrium total effort level may
not be maximized in the even contest.
Ability Parameter
I examine how the equilibrium effort ratio, the prize dissipation
ratios, and the players' equilibrium effort levels respond when
replacing one specific form of the function f with another. I do this
using f([theta]) = [sigma][sigma]/([sigma][sigma] + 1), where [sigma]
> 0. As explained in Section III, the parameter o represents player
1's abilities to convert effort into probability of winning
relative to player 2's. Note that different values of the ability
parameter [sigma] yield different specific forms of the function f.
Therefore, in this case, examining how those outcomes respond when
replacing one specific form of the function f with another amounts to
examining that when changing the value of the ability parameter o from
one to another. Indeed, what I do in this subsection is to examine the
effects of changing the value of the ability parameter [sigma] on the
equilibrium effort ratio, the prize dissipation ratios, and the
players' equilibrium effort levels.
PROPOSITION 4. (a) The equilibrium effort ratio,
[x*.sub.1]/[x*.sub.2], remains unchanged, regardless of the value of the
ability parameter [sigma]. (b) The prize dissipation ratios,
[x*.sub.1]/[v.sub.1] and [x*.sub.2]/[v.sub.2], are maximized in the even
contest, which occurs at [sigma] = [v.sub.2]/ [v.sub.1], and decrease as
the ability parameter [sigma] increases (decreases)from
[v.sub.2]/[v.sub.1]. (c) Each player's equilibrium effort level and
the equilibrium total effort level are also maximized at [sigma] =
[v.sub.1]/[v.sub.1]--that is, in the even contest and they decrease as
the ability parameter [sigma] increases (decreases) from
[v.sub.2]/[v.sub.1].
Using f([theta]) = [sigma][theta]/([sigma][theta] + 1)--in this
case, the players' contest success functions are [p.sub.1] =
[sigma][x.sub.1]/ ([sigma][x.sub.1] + [x.sub.2]) and [p.sub.2] =
[x.sub.2]([sigma][x.sub.1]+ [x.sub.2])--we obtain [x*.sub.1] =
[sigma][v.sup.2.sub.1][v.sub.2]/[sigma] [v.sub.1] + [v.sub.2],
[x*.sub.2] = [sigma][v.sub.1][v.sup.2.sub.2]/ [sigma][v.sub.1] +
[v.sub.2]).sup.2] and [x*.sub.1] + [x*.sub.2] =
[sigma][v.sub.1][v.sub.2]([v.sub.1] + [v.sub.2])/([[sigma][v.sub.1] +
[v.sub.2]).sup.2]. Given these expressions, it is straightforward to
obtain Proposition 4. I can also obtain part (a) directly from part (c)
of Proposition 1. To understand part (b) properly, recall from
Proposition 1 that the prize dissipation ratios for the players are the
same whatever specific form of the function f I use, and recall that the
even contest is defined as a contest in which both players have the same
probability of winning at the Nash equilibrium. Baik (1994) shows that
in this special case, the even contest occurs when the valuation ratio,
[v.sub.1]/[v.sub.2], times the ability parameter [sigma] equals unity.
This means that, given [v.sub.1] and [v.sub.2], the even contest occurs
at ([sigma] = [v.sub.2]/[v.sub.1].
Part (a) implies that the equilibrium effort ratio depends solely
on the players' valuations for the prize. Each player's
equilibrium effort level depends not only on the players'
valuations for the prize, but also on the ability parameter
[sigma]--that is, their abilities. However, because the players'
equilibrium effort levels are "equally" influenced by the
players' abilities, their ratio specifically, the equilibrium
effort ratio--does not depend on the players' abilities.
Parts (b) and (c) say that the prize dissipation ratio for each
player, each player's equilibrium effort level, and the equilibrium
total effort level are maximized in the even contest. An intuitive
explanation for this follows. Here the winner is determined by the
players' valuations for the prize and their abilities
together--namely, by their composite strength. I then believe that, when
both players have equal composite strength, competition for the prize is
the fiercest, and therefore the players' equilibrium effort levels
are maximized. Now note that the even contest is just the one in which
both players have equal composite strength: The valuation ratio,
[v.sub.1]/[v.sub.2], times the ability parameter [sigma] equals unity.
Parts (b) and (c) also say that starting from the even contest, as
player i becomes stronger relative to player j, both players exert less
effort and thus the equilibrium total effort level becomes smaller. (13)
An intuitive, interesting explanation for this result is that player
i--the player who becomes stronger relative to the opponent--eases up
facing an opponent weaker than before, whereas player j gives up, facing
an opponent stronger than before.
I can also obtain Proposition 4 by using Figure 1. In the case
where f([theta]) = [sigma][theta]/ ([sigma][theta] + 1), the payoff
function for player 1 is [[pi].sub.1] =
{[sigma][x.sub.1]/([sigma][x.sub.1] + [x.sub.2])} - [x.sub.1], and that
for player 2 is [[pi].sub.2] = [v.sub.2]{[x.sub.2]/ ([sigma][x.sub.1] +
[x.sub.2])} - [x.sub.2]. Because player 1's payoff function can be
rewritten as [[pi].sub.1] = [[sigma][v.sub.1]/([sigma][x.sub.1] +
[x.sub.2])} - [sigma][x.sub.1]/[sigma] maximizing player 1's
expected payoff over his or her effort level is equivalent to maximizing
[[phi].sub.1] = [sigma][v.sub.1]{[sigma][x.sub.1]/ ([sigma][x.sub.1] +
[x.sub.2])} - [sigma][x.sub.1] over his or her effort level. This means
that the players' equilibrium effort levels in the original contest
are the same as those obtained from a new contest in which the payoff
function for player 1 is [[phi].sub.1] =
[sigma][v.sub.1]{[sigma].sub.1]/([sigma][x.sub.1] + x2)} -
[sigma][x.sub.1], and that for player 2 is [[pi].sub.2] = [v.sub.2]
{[x.sub.2]/([sigma][x.sub.1] + [x.sub.2])} - [x.sub.2]. Let [w.sub.1] =
[sigma][v.sub.1] and let [y.sub.1] [equivalent to] [sigma][x.sub.1].
Then I have [[phi].sub.1] = [w.sub.1]{[y.sub.1]/([y.sub.1] + [x.sub.2])}
- [y.sub.1] and [[pi].sub.2] = [v.sub.2]{[x.sub.2]/([y.sub.1] +
[x.sub.2])} - [x.sub.2]. This is exactly the contest underlying Figure
1. All I need to change in the figure is to replace [v.sub.1]/[v.sub.2]
and [x*.sub.1]/[x*.sub.2] with [w.sub.1]/[v.sub.2] and
[y*.sub.1]/[x*.sub.2], respectively. The prize dissipation ratio for
each player in the new contest is identically equal to that in the
original contest: [y*.sub.1]/[w.sub.1] [equivalent to]
[x*.sub.1]][v.sub.1] and [x*.sub.2]/[v.sub.2] [equivalent to]
[x*.sub.2]/[v.sub.2]. Now note that a change in the valuation ratio in
the new contest comes solely from a change in the value of the ability
parameter [sigma] in the original contest. Therefore, I can obtain
Proposition 4 from the results illustrated in Figure 1. For example, the
result in part (b) that the prize dissipation ratios are maximized when
[sigma = [v.sub.2]/[v.sub.1] can be obtained from the
result--illustrated in the figure--that they are maximized when
[w.sub.1]/[v.sub.2] = 1.
V. CONCLUSIONS
I have considered a contest in which two asymmetric players compete
with each other by expending irreversible effort to win a prize, and
each player's probability of winning is a function of the ratio of
the two players' effort levels. Asymmetries between the players
arise because of the different valuations for the prize or their
different abilities to convert effort into probability of winning or
both. First, in Section III, I characterized the Nash equilibrium of the
simultaneous-move game and examined the equilibrium effort ratio, the
prize dissipation ratios, and the players' equilibrium effort
levels. Then, in Section IV, I examined how the equilibrium effort
ratio, the prize dissipation ratios, and the players' equilibrium
effort levels respond when the players' valuations change or when
the function f takes a different specific form.
In Section III, I found the following. The equilibrium effort ratio
is equal to the valuation ratio, and that the prize dissipation ratios
for the players are the same. These results hold for any specific form
of the function f. The equilibrium effort ratio does not depend on the
function f, whereas the prize dissipation ratio for each player does.
Finally, the prize dissipation ratio for each player is less than or
equal to the minimum of the players' probabilities of winning at
the Nash equilibrium and thus never exceeds a half.
In Section IV, when the players' valuations for the prize
change, I found the following. If the players' valuations for the
prize change proportionately and thus the valuation ratio remains
unchanged, then the equilibrium effort ratio and the prize dissipation
ratios remain unchanged, but each player's equilibrium effort level
changes in the same direction as the players' valuations do. If the
valuation ratio changes, then the equilibrium effort ratio changes in
the same direction as the valuation ratio does, but the prize
dissipation ratios change in either the same or the opposite direction.
Examining the effects of changing the players' abilities on the
equilibrium effort ratio, the prize dissipation ratios, and the
players' equilibrium effort levels, I found the following. The
equilibrium effort ratio depends solely on the players' valuations
for the prize it does not depend on the players' abilities. The
prize dissipation ratio for each player, each player's equilibrium
effort level, and the equilibrium total effort
level are maximized in the even contest.
Can I extend this two-player model to an n-player one? What do
ratio-form contest success functions look like in n-player asymmetric
contests? Are the prize dissipation ratios for the players the same in
n-player asymmetric contests? Does the prize dissipation ratio for each
player never exceed 1/n in n-player asymmetric contests? These are
interesting questions. I leave them for future research.
REFERENCES
Appelbaum, E., and E. Katz. "Seeking Rents by Setting Rents:
The Political Economy of Rent Seeking." Economic Journal, 97(387),
1987, 685-99.
Balk, K. H. "Effort Levels in Contests with Two Asymmetric
Players." Southern Economic Journal, 61(2), 1994, 367-78.
--. "Difference-Form Contest Success Functions and Effort
Levels in Contests." European Journal of Political Economy, 14(4),
1998, 685-701.
--. "Rent-Seeking Firms, Consumer Groups, and the Social Costs
of Monopoly." Economic Inquiry, 37(3), 1999, 541-53.
Baik, K. H., and S. Lee. "Strategic Groups and Rent
Dissipation." Economic Inquiry, 39(4), 2001, 672-84.
Che, Y.-K., and I. L. Gale. "Caps on Political Lobbying."
American Economic" Review, 88(3), 1998, 643-51.
Clark, D. J., and C. Riis. "Competition over More Than One
Prize." American Economic Review, 88(1), 1998, 276-89.
Dixit, A. "Strategic Behavior in Contests." American
Economic Review, 77(5), 1987, 891-98.
Ellingsen, T. "Strategic Buyers and the Social Cost of
Monopoly." American Economic Review, 81(3), 1991, 648-57.
Friedman, J. W. Game Theory with Applications to Economics, 2nd ed.
New York: Oxford University Press, 1990.
Hillman, A. L., and J. G. Riley. "Politically Contestable
Rents and Transfers." Economics and Politics, 1(1), 1989, 17-39.
Hirshleifer, J. "Conflict and Rent-Seeking Success Functions:
Ratio vs. Difference Models of Relative Success." Public Choice,
63(2), 1989, 101-12.
Hurley, T. M., and J. F. Shogren. "Effort Levels in a Cournot
Nash Contest with Asymmetric Information." Journal of Public
Economics, 69(2), 1998, 195-210.
Konrad, K. A. "Trade Contests." Journal of International
Economics, 51(2), 2000, 317-34.
Kooreman, P., and L. Schoonbeek. "The Specification of the
Probability Functions in Tullock's Rent-Seeking Contest."
Economics Letters, 56(1), 1997, 59-61.
Krishna, V., and J. Morgan. "An Analysis of the War of
Attrition and the All-Pay Auction." Journal of Economic Theory,
72(2), 1997, 343-62.
Lee, T., and L. L. Wilde. "Market Structure and Innovation: A
Reformulation." Quarterly Journal of Economics, 94(2), 1980,
429-36.
Leininger, W. "More Efficient Rent-Seeking--A Munchhausen
Solution." Public Choice, 75(1), 1993, 43-62.
Loury, G. C. "Market Structure and Innovation." Quarterly
Journal of Economics, 93(3), 1979, 395-410.
Nitzan, S. "Collective Rent Dissipation." Economic
Journal, 101(409), 1991, 1522-34.
--. "Modelling Rent-Seeking Contests." European Journal
of Political Economy, 10(1), 1994, 41-60.
Nti, K. O. "Comparative Statics for Contests and Rent-Seeking
Games." International Economic Review, 38(1), 1997, 43-59.
--. "Rent-Seeking with Asymmetric Valuations" Public
Choice, 98(3-4), 1999, 415-30.
Rogerson, W. P. "The Social Costs of Monopoly and Regulation:
A Game-Theoretic Analysis." Bell Journal of Economics, 13(2), 1982,
391-401.
Rosen, S. "Prizes and Incentives in Elimination
Tournaments." American Economic Review, 76(4), 1986, 701-15.
Skaperdas, S. "Contest Success Functions." Economic
Theory, 7(2), 1996, 283-90.
Szidarovszky, F., and K. Okuguchi. "On the Existence and
Uniqueness of Pure Nash Equilibrium in Rent-Seeking Games." Games
and Economic Behavior, 18(1), 1997, 135-40.
Tullock, G. "Efficient Rent Seeking," in Toward a Theory
of the Rent-Seeking Society, edited by J. M. Buchanan, R. D. Tollison,
and G. Tullock. College Station, TX: Texas A&M University Press,
1980, 97-112.
(1.) The contest success function for a player is the rule that
describes the relationship between effort levels of the players and his
or her probability of winning. Several forms of contest success
functions have been used in the literature on the theory of contests.
See Baik (1998, 686-87) for detailed explanations of different forms of
contest success functions.
(2.) This example may be rather contrived, but is useful in
understanding the model and results of this article.
(3.) Alternatively, the government can design and use a different
winner-selecting mechanism in which the communities face more general
ratio-form contest success functions.
(4.) These contest success functions hold for [x.sub.2] >0. When
[x.sub.j] = 0, one may assume that [p.sub.i] = 1 and [p.sub.j] = 0 for
[x.sub.i]>0, and [p.sub.1] [p.sub.2] = 1/2 for [x.sub.i] = 0.
(Throughout the article, when I use i and j at the same time, I mean
that i [not equal to] j.) The ratio-form contest success functions are
homogeneous of degree zero in the players' effort levels.
Hirshleifer (1989) coined the term ratio-form contest success function.
He called the following pair of logit-form contest success
functions--which was first used in Tullock (1980)--the ratio-form
contest success functions: [p.sub.1] = [x.sup.r.sub.1]/([x.sup.r.sub.1]
+ [x.sup.r.sub.2]) and [p.sub.2] = [x.sup.r.sub.2]/([x.sup.r.sub.1] +
[x.sup.r.sub.2]), where the exponent r is a positive constant.
Hirshleifer (1989) also coined the term difference-form contest success
function. His difference-form contest success functions are just the
logistic-form ones: [p.sub.1]([x.sub.1], [x.sub.2]) = 1/[1 +
exp{k([x.sub.j] - [x.sub.i])}] for i,j = 1,2 where k is a positive
constant. Note that given these difference-form contest success
functions, player i's probability of winning is not zero even if he
or she expends zero effort. Hirshleifer (1989) shows that unlike in a
model with the ratio-form contest success functions, one-sided
submission or two-sided peace can occur in equilibrium in a model with
the difference-form contest success functions. He argues that the
ratio-form contest success functions are applicable to, for example,
military struggles between two nations in which there are
"idealized" conditions, such as an undifferentiated battlefield and full information, while the difference-form contest
success functions are applicable to, for example, military struggles
with sanctuaries, refuges, and/or imperfect information.
(5.) Hillman and Riley (1989) classify contests according to
whether the player who expends the largest effort level wins the prize
with certainty. Their classification divides contests into perfectly
discriminating contests and imperfectly discriminating contests. The
player who expends the largest effort level wins the prize with
certainty in a perfectly discriminating contest, but may lose it in an
imperfectly discriminating contest. According to their classification,
contests under consideration in this article--that is, contests with
ratio-form contest success functions--are imperfectly discriminating
ones. Other examples of imperfectly discriminating contests include
contests with logit-form, probit-form, or difference-form contest
success functions. On the other hand, first-price all-pay and
second-price all-pay contests belong to perfectly discriminating
contests (see Baik, 1998).
(6.) There may exist some real-world situations in which a
player's probability of winning is increasing in his or her own
effort level at an increasing rate for a while and then increasing at a
decreasing rate. Even in such situations, however, an equilibrium occurs
only where each player's probability of winning is increasing at a
decreasing rate.
(7.) See, for example, Tullock (1980), Rogerson (1982), Appelbaum
and Katz (1987), Hillman and Riley (1989), Ellingsen (1991), Nitzan
(1991), Leininger (1993), Baik (1994, 1999), and Baik and Lee (2001).
(8.) In the last two examples, the exponent r is a positive
constant but, to satisfy Assumption 1, it must be
"small"--roughly, it must be less than 2. Other specific forms
of the function f are f([theta]) = 1n(1 + [theta])/{1 + 1n(1 + [theta])}
and f([theta]) [{[theta]/(1 + [theta])}.sup.r], where 0<r [less than
or equal to] 1.
(9.) If I adopt the assumption mentioned in footnote 4, then, when
player j expends zero effort, player i's best response is to expend
infinitesimally small effort.
(10.) Let J([x.sub.1], [x.sub.2]) be the Jacobian of equations (3)
and
(4). Then
[MATHEMATICAL CONVERSION NOT REPRODUCIBLE IN ASCII.]
If J([x.sub.1], [x.sub.2]) is negative quasi-definite for all
strategy profiles, then the game has a unique Nash equilibrium (see
Friedman, 1990, 86). (Note that this uniqueness theorem gives
sufficient, but not necessary, conditions for uniqueness.) It is
straightforward to show that J([x.sub.1], [x.sub.2]) is negative
quasi-definite for all strategy profiles if [1 + 2[square root of
[v.sub.1] [v.sub.2]/([v.sub.1] + [v.sub.2])] f'([theta]) +
[theta]f"([theta]) >0 and [1 - 2[square root of [v.sub.1]
[v.sub.2]/([v.sub.1] + [v.sub.2])] f'([theta]) + [theta]f ([theta])
<0 for all [theta] in [R.sub.+]. Szidarovszky and Okuguchi (1997)
prove the existence and uniqueness of a pure-strategy Nash equilibrium
in rent-seeking contests with logit-form contest success functions.
(11.) Nti (1999) also obtains this result in a rent-seeking contest
with Tullock's pair of contest success functions.
(12.) Player i's best response to player j's
"sufficiently high" effort level may be to exert zero effort
and may remain unchanged at the zero effort levels, when the
players' valuations for the prize change.
(13.) An increase in the ability parameter [sigma] means that
player 1 becomes stronger relative to player 2. Naturally, it also means
that player 2 becomes weaker relative to player 1.
KYUNG HWAN BAIK, I am grateful to Dennis W. Jansen, In-Gyu Kim,
Alain Marciano, Jay Shogren, Cheng-Chen Yang, and two anonymous referees
for their helpful comments and suggestions. An earlier version of this
article was presented at the 76th Annual Conference of the Western
Economic Association International, San Francisco, CA, July 2001. This
work was supported by the Brain Korea 21 Project in 2001.
Baik: Professor, Department of Economics, Sungkyunkwan University,
Seoul 110-745, South Korea. Phone 82-2-760-0432, Fax 82-2-744-5717,
E-mail khbaik@skku.edu