Rent-Seeking Contests with Incomplete

Information
Mark Fey∗

Department of Political Science

University of Rochester

September, 2007

Abstract

We consider rent-seeking contests with two players that each have

private information about their own cost of effort. We consider both
discrete and continuous distributions of costs and give results for each
case, focusing on existence of equilibria.

JEL Classification: D72; C72

Keywords: rent-seeking, contests, conflict, private information, equi-

librium existence

∗
Harkness Hall, University of Rochester, Rochester, NY, 14618. Office: 585-275-5810.
Fax: 585-271-1616. Email: markfey@mail.rochester.edu.
1 Introduction
A rent-seeking contest is a situation in which players expend costly effort
to gain a reward. Many conflict situations can be described by rent-seeking
contests, including political campaigns, patent races, war fighting, lobbying
efforts, labor market competition, legal battles, and professional sports. The
reward in a rent-seeking contest may be indivisible, such as electoral office
or a patent right, or it may be divisible, such as market share or vote share.
In the former case, expending more effort increases the probability that a
player will win the prize. In the latter case, expending more effort increases
the share of the prize.
An important vehicle for investigating the logic of rent-seeking contests is
the model of Tullock (1980). This work has spawned a large literature, some
of which is surveyed in Lockard and Tullock (2001) and Corchón (2007). In
this paper, we contribute to this literature by developing a model of rent-
seeking contests in which players have incomplete information about the cost
of effort to other players. That is, players are uncertain about the value of
others for expending effort.1
We consider rent-seeking contests with two players that each have private
information about their own cost of effort. We model such a contest as
Bayesian game in which each player’s cost is drawn from a distribution of
possible costs. We consider both discrete and continuous distributions of
costs and give results for each case. We focus on existence of equilibria,
which, in the continuous case, requires the use of a functional analysis fixed
point argument.
The focus on rent-seeking contests with two-sided incomplete informa-
tion and continuous cost distributions is what sets this paper apart from
other papers that consider asymmetric information in contests. For exam-
1
It is important to distinguish this informational type of uncertainty from the uncer-
tainty due to risk aversion that has been examined in the literature (Hillman and Katz,
1984; Van Long and Vousden, 1987).

1
ple, Harstad (1995) looks at a model with asymmetrically informed agents,
but in the context of a winner-take-all game. Hurley and Shogren (1998b)
and Schoonbeek and Winkel (2006) both consider models with one-sided in-
complete information, while Hurley and Shogren (1998a) and Malueg and
Yates (2004) present models with two-sided incomplete information in which
both players’ valuations are either high or low. Additionally, recent work
on existence of equilibria in rent-seeking contests includes Cornes and Hart-
ley (2005) and Malueg and Yates (2006), but these papers assume complete
information. Bernardo et al. (2000) embed a rent-seeking contest with one-
sided incomplete information in a moral hazard model of litigation. Finally,
Wärneryd (2003) gives a Bayesian model of rent-seeking contests in which
there is asymmetric information about the common value of the prize. Al-
though we deal with the private value case of individual cost of effort, some
of the techniques of analysis are similar.
The paper is organized in the following fashion. The next section contains
the definitions and notation of the model, as well as a summary of the results
for the complete information setting. In section 3, we examine a simple
example of incomplete information in which each player has only two possible
costs of effort. We explicitly solve for the equilibrium and provide some
comparative statics. In section 4, we consider the continuous case, with
a uniform distribution of possible costs. The main result is an existence
proof. We also graphically present a numerical estimate of the equilibrium
and provide some comparative statics and properties for this case. Finally,
section 5 concludes.

2 Notation and Definitions

We model a rent-seeking contest as a game between two players who must
choose a level of costly effort to obtain a share of a prize. Each player chooses
a level of effort ei ≥ 0 for i = 1, 2. The effort level of both players is denoted

2
e = (e1 , e2 ). The effort levels chosen by the players determine a portion
fi (e) ∈ [0, 1] for player i. We interpret a “portion” in one of two ways —
either as a fraction of a divisible good or as the probability of receiving a
nondivisible good. By the expected utility theorem, these interpretations are
mathematically equivalent if both players are risk neutral.2 We assume that
the benefit a player receives is linear in the player’s portion and the cost a
player pays is linear in effort. That is, there are constants αi , βi , γi , and δi
with βi > 0 and δi > 0, such that player i’s expected utility is

ui (e) = αi + βi fi (e) − (γi + δei ).

Without loss of generality, we can take αi − γi = 0. Likewise, we can divide

by βi , which gives
δi
ui (e) = fi (e) − ei .
βi
To operationalize the analysis, we specify the following functional form for
fi (e): 
e / P e if
P
i j j j ej > 0
fi (e) =
1/2 otherwise.

This functional form is common in the literature (Tullock, 1980; Nitzan, 1991;
Szidarovsky and Okuguchi, 1997) and is a simple form of the contest suc-
cess functions axiomatized by Skaperdas (1996) and Clark and Riis (1998).3
Letting ci = δi /βi , the expected utility for player i is

ei
ui (e) = P − e i ci ,
j ej

unless ei = 0, in which case ui (e) = 0. By defining ci in this way, we include

the case in which the individuals’ valuations of the prize (the βi term) differ
2
This is not true if the player are risk averse (Hillman and Katz, 1984; Van Long and
Vousden, 1987).
3
In fact, our results do not depend on the value of fi (e) at e = (0, 0).

3
(Hillman and Riley, 1989).
If both players have complete information about all aspects of the game,
the equilibrium of this game is easily found (and well known). If the players
have the same cost of effort, c, then the first order condition for player i is

e
P −i 2 = c. (1)
( j ej )
As this must hold for both players, it clear that, in equilibrium, e1 = e2 .4
Substituting this back into equation (1) gives the solution e1 = e2 = 1/(4c)
in equilibrium. As both players would be strictly better off by choosing
e1 = e2 = 0, this equilibrium solution is inefficent, again, a standard result.
Building on this complete information setup, in this paper we model a
contest as a game of incomplete information in which players are uncertain
about the other’s cost of effort.5 In the sections that follow, we consider both
the discrete and continuous cases. We introduce some general notation here
and give precise details in the later sections.
We suppose that each player’s cost is drawn independently from a distri-
bution G(c) before the game is played. G(c) may be discrete or continuous
and we denote the support of G by C ⊆ R+ . Each player’s own cost is private
information and thus G represents the players’ common prior of the type of
the other, where we identify the type of a player with their cost. A (pure)
strategy for player i is a function σi : C → R+ . In other words, a player i
with cost ci chooses an effort level ei = σi (ci ).
The equilibrium concept we use for this simultaneous Bayesian game
is Bayesian-Nash equilibrium. Specifically, a strategy profile (σ1∗ , σ2∗ ) is a
Bayesian-Nash equilibrium if, for every c1 ∈ C,
Z Z
u1 (σ1∗ (c1 ), σ2∗ (c)) dG(c) ≥ u1 (σ10 (c1 ), σ2∗ (c)) dG(c)
C C
4
It is easy to check that ei > 0 holds for both players in equilibrium.
5
See Gradstein (1995) for a model with variable effort effectiveness instead of variable
costs.

4
for every σ10 , and similarly for player 2. We restrict our attention to symmetric
equilibria. These are equilibria in which players of the same type take the
same action. In our setting, a symmetric Bayesian-Nash equilibrium satisfies
σ1∗ = σ2∗ = σ ∗ .

3 Results: Discrete Case

In this section, we analyze the simplest possible formulation of this game
of incomplete information; namely, the discrete case with only two types
for each player. This simple formulation of uncertainty has been previously
examined by Hurley and Shogren (1998a) and Malueg and Yates (2004).
In particular, we assume that C = {cL , cH }, with cH > cL > 0, and that
each player i (independently) draws cost ci = cL with probability 1/2 and
draws cost ci = cH with probability 1/2. Thus, a (symmetric) strategy is a
function σ : C → R+ . As the domain of this function has only two elements,
it is sufficient to consider two values, eL and eH , defined by eL = σ(cL ) and
eH = σ(cH ).
The first order condition is given by

1 eH 1 eL
( ) + ( ) = ci . (2)
2 (ei + eH )2 2 (ei + eL )2

When ci = cL , ei = eL must satisfy this condition. This gives

eH 1
2
+ = 2cL . (3)
(eL + eH ) 4eL

Likewise, when ci = cH , ei = eH must satisfy the first order condition,6 so

1 eL
+ = 2cH . (4)
4eH (eL + eH )2
6
It is easy to show that eL and eH must both be positive.

5
Manipulating equations 3 and 4 gives

4eL eH + (eL + eH )2 = 2cL (4eL )(eL + eH )2 (5)

and
(eL + eH )2 + 4eH eL = 2cH (4eH )(eL + eH )2 . (6)

From equations (5) and (6), it is clear that cL eL = cH eH must hold. Equiv-
alently, we must have
eH cL
= . (7)
eL cH
In other words, the ratio of the equilibrium effort levels must be the inverse
of the ratio of the costs.7 Plugging eH = eL (cL /cH ) into equation (5) yields

4 ccHL + (1 + cL 2
cH
)
eL = cL 2 , (8)
8cL (1 + cH
)

and thus by equation (7),

4 ccHL + (1 + cL 2
cH
)
eH = cL 2 . (9)
8cH (1 + cH
)

To summarize, in this simple discrete case, a unique (symmetric) equilibrium

exists.
We now turn to analyzing the comparative statics of this case. It is
straightforward but tedious to show that ∂eL /∂cL < 0. Thus, holding the
high cost constant, increasing the low cost decreases the effort level of a low
type. Perhaps a more interesting comparative static is the case in which both
costs change in such a way that ccHL is held constant. Then it is immediate
from equations (8) and (9) that the effort levels of both types of player change
by an inverse amount. Thus, when both costs are doubled, for example, the
effort levels of both types of player decrease by half.
7
See Malueg and Yates (2004) for a similar result.

6
 Finally, consider what happens to the equilibrium as cL converges to cH .
From equation (7), it is clear that eL goes to eH as cL goes to cH . Moreover,
if cL and cH converge to some value c, then eL and eH must converge to 1/4c,
which is the solution to the complete information version of the model.

4 Results: Continuous Case

In this section we prove existence and numerically analyze the equilibrium
for the case of a continuous distribution of costs. By considering the case in
which each player has a continuum of possible types, we allow for a much
richer and more general specification of uncertainty than previous models
which consider only two possible types of each player, such as Hurley and
Shogren (1998a) and Malueg and Yates (2004).
For simplicity, we suppose that each player’s cost ci is drawn indepen-
dently from a uniform distribution on an interval of length one, that is
G(c) ∼ U [c, c + 1], with c > 0 and C = [c, c + 1]. With this assump-
tion, the expected utility for player i with cost ci , when the opponent plays
strategy σ, is given by

ei
Z
ui (ei , σ; ci ) = − ei ci + dt. (10)
C ei + σ(t)

Because player i can guarantee a payoff of at least zero by setting ei = 0,

the optimal choice of ei must satisfy ui (ei , σ; ci ) ≥ 0 for all ci ∈ C. As the
integral in equation (10) is bounded by 1, this implies that the optimal choice
of ei must satisfy 0 ≤ ei ≤ 1/ci . On the interval C = [c, c + 1], then, an
optimal choice of ei must satisfy 0 ≤ ei ≤ 1/c. We denote this upper bound
by E = 1/c.
A player with cost ci faces an optimization problem of maximizing equa-
tion (10) subject to ei ≥ 0. The Kuhn-Tucker conditions for a given value of

7
ci are the following:

σ(t)
Z
− ci + 2
dt = 0 for ei > 0, and (11)
C (ei + σ(t))
σ(t)
Z
− ci + 2
dt ≤ 0 for ei = 0. (12)
C (ei + σ(t))

Examining the integral in equation (11), it is clear that it is strictly decreasing

in ei and goes to zero as ei gets large. Thus, it follows that this first order
condition is satisfied by exactly one value of ei . A best response for player i
to σ, which we denote bσ (c) or b(c), is precisely the value of ei that satisfies
the first order condition with cost c. Thus, the best response function bσ (c)
is well-defined, nonnegative, and decreasing in c.8 In addition, it is an easy
consequence of the implicit function theorem that the best response function
is continuous.
We now present the main result of this section—that a equilibrium exists
in our model.

Theorem 1 If G(ci ) ∼ U [c, c+1], with c > 0, for i = 1, 2, then a symmetric

pure strategy Bayesian-Nash equilibrium exists.

The proof of this theorem is contained in the appendix. As is standard,

the method of proof of existence is to find a fixed point of the best response
mapping. However, this mapping is defined on a subset of a space of functions
which, like most such sets, is not compact (under the standard sup norm).
Therefore we use an alternative approach. We have already established that
the best response functions are monotone and continuous. In the appendix,
we show that the set of best response functions is equicontinuous and that
the operator that maps the opponent’s strategy into a best response function
is continuous. This allows us to establish existence using Schauder’s Fixed
Point Theorem. This approach is more flexible than a contraction mapping
8
In fact, it is clear that b(c) is strictly decreasing for values of c with b(c) > 0.

8
 Effort
5

Cost
0.2 0.4 0.6 0.8 1

Figure 1. Numerical solution and comparison to complete information

argument (which only requires completeness), but it does not insure unique-
ness of the fixed point. We conjecture that the equilibrium to the model is,
in fact, unique, but leave a proof for future work.
Theorem 1 shows that an equilibrium exists. To find a closed-form so-
lution for it, we must solve the integral equation given in equation (11) for
all c ∈ C. One approach to accomplish this is by working with the inverse
of the solution, which transforms the integral equation from nonlinear into
linear form, but such an approach appears to be intractable in this case.
On the other hand, it is possible to numerically calculate a solution by a
standard iterative method. For the case of c = .01, the numerical solution is
shown in Figure 1. In the figure, the horizontal axis is the cost value drawn
by a player and the vertical axis is the equilibrium effort expended. The
numerical solution is given by the solid line in the figure. This solution was
calculated numerically over a grid of 100 elements and thus the figure is a
good approximation of the true equilibrium strategy.
Figure 1 also illustrates how the equilibrium effort level under incomplete
information compares to the effort level of the complete information symmet-
ric game from section 2. Under complete information, the equilibrium effort
level is ei = 1/(4c), which is indicated by the dashed line in Figure 1. It is
interesting to observe that the two functions are qualitatively similar, but

9
the incomplete information strategy always calls for slightly less effort than
the complete information version.9 This result echoes an earlier finding by
Hurley and Shogren (1998a) that, in a model in which each player could have
either a high or low valuation of the prize, asymmetric information makes ef-
fort more risky, which tends to decrease effort levels. Figure 1 illustrates that
this effect carries over to a model in which both players have a continuum of
possible types.
Finally, inspection of Figure 1 shows that both players always exert posi-
tive effort in equilibrium. In fact, this is true for any symmetric equilibrium
of this game. To see this, suppose that there is a symmetric equilibrium
in which a non-zero interval of costs gives zero effort. Let the probability
that such a cost is realized be p0 > 0. If this occurs, the other player can
gain the prize with certainty by exerting an arbitrarily small amount of ef-
fort. In particular, a level of effort that satisfies 0 < ei < p0 /ci yields a
gain in expected utility. With this choice of effort at cost ci , a player does
strictly better than with ei = 0. Thus, there can be no interval of costs for
which zero effort is optimal in a symmetric equilibrium. Moreover, a similar
argument shows that this result continues to hold in any symmetric equilib-
rium of a game with n > 2 players.10 This conclusion differs from work on
entry in complete information rent-seeking contests in which some players
choose not to participate by choosing zero effort (Gradstein, 1995; Higgins
et al., 1985).11 However, such situations under complete information can
occur only in “asymmetric” contests in which players have different valua-
tions. Our finding that players always exert positive effort under incomplete
information applies to symmetric equilibria of a symmetric model (i.e., all
players have an identical cost distribution). Thus, even though players can
9
Of course, the dashed line is not a strategy in the same way that the solid line is, but
rather describes equilibrium effort as the cost of effort in the complete information game
is varied.
10
The only modification required in the above argument is that the level of effort must
satisfy 0 < ei < p0n−1 /ci in a game with n players.
11
But see Wärneryd (2006) for a similar result in a model with asymmetric information.

10
have different realized costs in our model, they are identical ex ante. These
differences in the symmetry of the models suggest that caution is warranted
in making comparisons between them.

5 Conclusion
In this paper, we have examined rent-seeking contests under various forms of
incomplete information. The analysis of these models has been complex in
places, but we have gained some insight into the role that information plays
in this class of games. Our main result is a proof of existence of equilibrium
when costs are drawn from a continuous distribution.
Much work remains to be done in this vein, particulary for the continu-
ous case. First, the model can be made more general by allowing for more
general distributions of costs and more than two players. Fortunately, it ap-
pears that in both cases, the line of proof given in the appendix should go
through with only minor modifications. However, numerical calculation of
the result would be more difficult. A second line of work involves generaliz-
ing the contest success function fi (e). It is likely this will be a more difficult
task to accomplish. Another aspect of the problem that remains open is
an examination of efficiency in the continuous case. Direct comparison with
the complete information equilibrium demonstrates that inefficiency is still
prevalent, but more work in this area remains to be done. Finally, we have
restricted our attention to symmetric equilibria in this paper. Future work
that considers asymmetric equilibria in this model and/or asymmetric models
of rent-seeking contests may be illuminating and ultimately more realistic.

11
6 Appendix: Proof of Existence
Theorem If G(ci ) ∼ U [c, c + 1], with c > 0, for i = 1, 2, then a symmetric
Bayesian-Nash equilibrium exists.

Proof : Let C[a, b] be the space of all real continuous functions on the interval
[a, b]. Equipped with the sup norm, kf (x)k = supx∈[a,b] | f (x) |, this space is
a Banach space. We are interested in the set of nonnegative continuous
functions on the interval C = [c, c + 1] bounded by E = 1/c. Formally, define

CE [c, c + 1] = {f ∈ C[c, c + 1] | kf k ≤ E and f (x) ≥ 0, x ∈ [c, c + 1]}. (13)

Clearly, this is a closed, convex subset of C[c, c + 1].12

For a given function σ ∈ CE , let the best response function bσ (c) be as
defined above. As bσ (c) is bounded by E and is continuous, bσ (c) ∈ CE .
Therefore we can define T : CE → CE to be the mapping σ 7→ bσ . As the
best response function is unique, this mapping is a function. An equilibrium
is a fixed point of this function, σ ∗ = T σ ∗ .
Because CE is not compact with the sup norm, we rely on a fixed point
theorem for compact operators. An operator S : X → Y , X and Y Banach
spaces, is compact if it is continuous and maps bounded sets into relatively
compact sets.13 We can now state the Schauder Fixed Point Theorem, a
useful discussion of which is found in Zeidler (1986).

Theorem (Schauder) Let X be a nonempty, closed, bounded, convex sub-

set of a Banach space and suppose S : X → X is a compact operator. Then
S has a fixed point.

In order to establish that T is a compact operator, the characterization of

relative compactness in the space of continuous functions given by the Arzelà-
12
For simplicity, in what follows we exclude from this set the constant zero function.
Excluding this element would mean this set is not closed. This issue can be resolved by
considering a sequence of subsets of CE requiring kf k ≥ ε, with ε → 0.
13
A set is relatively compact if its closure is compact.

12
Ascoli Theorem is very useful. This theorem uses the following two defini-
tions. A set of (real-valued) functions F is equicontinuous on the interval C if
for every ε > 0 there is a δ > 0 such that | x−y | < δ implies | f (x)−f (y) | < ε
for all f ∈ F and all x, y ∈ C. A set of (real-valued) functions F is uniformly
bounded on the interval C if there is a value M < ∞ such that | f (x) | ≤ M
for all f ∈ F and all x ∈ C.

Theorem (Arzelà-Ascoli) A set of functions in C[a, b], with the sup norm,
is relatively compact if and only if it is uniformly bounded and equicontinuous
on [a, b].

Clearly CE is uniformly bounded, so to establish our result we must show

that T is continuous and has equicontinuous range.
We first show that the range of T is a set of equicontinuous functions.
From the first order condition, we have

σ(t) σ(t)
Z Z
c1 − c 2 = dt − dt
(b(c1 ) + σ(t))2 C (b(c2 ) + σ(t))
2
ZC
σ(t) σ(t)
= 2
− dt
C (b(c1 ) + σ(t)) (b(c2 ) + σ(t))2
(b(c2 ) + σ(t))2 − (b(c1 ) + σ(t))2
Z
= σ(t) dt
C (b(c1 ) + σ(t))2 (b(c2 ) + σ(t))2
b2 (c2 ) − b2 (c1 ) + 2σ(t)(b(c2 ) − b(c1 ))
Z
= σ(t) dt
C (b(c1 ) + σ(t))2 (b(c2 ) + σ(t))2
b(c2 ) + b(c1 ) + 2σ(t)
Z
 
= b(c2 ) − b(c1 ) σ(t) dt
C (b(c1 ) + σ(t))2 (b(c2 ) + σ(t))2
b(c1 ) + σ(t) + b(c2 ) + σ(t)
Z
 
= b(c2 ) − b(c1 ) σ(t) dt
C (b(c1 ) + σ(t))2 (b(c2 ) + σ(t))2
Z
  σ(t)
= b(c2 ) − b(c1 ) 2
C (b(c1 ) + σ(t))(b(c2 ) + σ(t))

σ(t)
+ dt .
(b(c1 ) + σ(t))2 (b(c2 ) + σ(t))

The two integrals on this last line are both positive. Using the first order

13
condition and the fact that b(c) and σ(t) are both bounded by E, we have

σ(t) σ(t) c2
Z Z
dt ≥ dt = (14)
C (b(c1 ) + σ(t))(b(c2 ) + σ(t))2 C 2E(b(c2 ) + σ(t)) 2 2E

A similar argument shows that the second integral in the expression above
is bounded below by c1 /2E. Taking absolute values, we have
c c1  c
2
| c1 − c2 | ≥ | b(c2 ) − b(c1 ) | + > | b(c2 ) − b(c1 ) | (15)
2E 2E E

Now, fix ε > 0 and let δ = ε(c/E). Then | c1 − c2 | < δ implies | b(c2 ) −
b(c1 ) |(c/E) < ε(c/E) or | b(c2 ) − b(c1 ) | < ε. As this holds for all best
response functions and all c1 , c2 ∈ C, this establishes equicontinuity.
We now show that the operator T is continuous. In other words, if σn → σ
and bσn = T σn and bσ = T σ, then bσn → bσ . Convergence in a function space
with the sup norm is equivalent to uniform convergence, so we must show
bσn converges uniformly to bσ . So take a sequence σn → σ and a point c ∈ C.
The first order condition yields

σn (t) σ(t)
Z Z
dt = c = dt. (16)
C (bσn (c) + σn (t))2 C (bσ (c) + σ(t))2

And thus

σn (t) σ(t)
Z
2
− dt = 0
C (bσn (c) + σn (t)) (bσ (c) + σ(t))2
σn (t)(b2σ (c) + 2bσ (c)σ(t) + σ 2 (t)) − σ(t)(b2σn (c) + 2bσn (c)σn (t) + σn2 (t))
Z
dt = 0
C (bσn (c) + σn (t))2 (bσ (c) + σ(t))2
σn (t)b2σ (c) − σ(t)b2σn (c) + 2σ(t)σn (t)(bσ (c) − bσn (c)) + σ(t)σn (t)(σ(t) − σn (t))
Z
dt = 0.
C (bσn (c) + σn (t))2 (bσ (c) + σ(t))2

We can rewrite the first two terms in the numerator as

σn (t)b2σ (c) − σ(t)b2σn (c) = σn (t)b2σ (c) − σ(t)b2σ (c) + σ(t)b2σ (c) − σ(t)b2σn (c)

14
 = b2σ (c)(σn (t) − σ(t)) + σ(t)(b2σ (c) − b2σn (c))
= b2σ (c)(σn (t) − σ(t)) + σ(t)(bσ (c) + bσn (c))(bσ (c) − bσn (c)).

This gives

σ(t)(bσ (c) + bσn (c))(bσ (c) − bσn (c)) + 2σ(t)σn (t)(bσ (c) − bσn (c))
Z
dt
C (bσn (c) + σn (t))2 (bσ (c) + σ(t))2
Z 2
bσ (c)(σn (t) − σ(t)) + σ(t)σn (t)(σ(t) − σn (t))
+ dt = 0.
C (bσn (c) + σn (t))2 (bσ (c) + σ(t))2

Setting D = (bσn (c) + σn (t))2 (bσ (c) + σ(t))2 , we have

σ(t)(bσ (c) + bσn (c)) + 2σ(t)σn (t)

Z
(bσ (c) − bσn (c)) dt
C D
σ(t) − σn (t) σ(t)σn (t)(σn (t) − σ(t))
Z Z
= b2σ (c) dt + dt.
C D C D

Taking the absolute value of both sides and noting that the integral on the
left side of the equation is positive, we have

σ(t)(bσ (c) + bσn (c)) + 2σ(t)σn (t)

Z
| bσ (c) − bσn (c) | dt
C D
 
σ(t) − σ (t) σ(t)σ (t)(σ (t) − σ(t))
Z Z
 2 n n n 
= bσ (c) dt + dt ,
C D C D

and so

σ(t)(bσ (c) + bσn (c)) + 2σ(t)σn (t)

Z
| bσ (c) − bσn (c) | dt
C D
Z  Z 
2
 σ(t) − σn (t)   σ(t)σn (t)(σn (t) − σ(t)) 
≤ bσ (c)  dt +  dt
C D C D
| σ(t) − σn (t) | σ(t)σn (t)| σ(t) − σn (t) |
Z Z
2
≤ bσ (c) dt + dt.
C D C D

Now, for a given ε, choose n large enough that | σ(t) − σn (t) | < ε. This

15
means
Z 
σ(t)(bσ (c) + bσn (c)) σ(t)σn (t)
Z
| bσ (c) − bσn (c) | dt + 2 dt
C D C D
1 σ(t)σn (t)
Z Z
2
≤E ε dt + ε dt.
C D C D
R σ(t)σn (t)
Dividing both sides by C D
dt yields
" σ(t)(bσ (c)+bσn (c))
# " #
E 2 C D1 dt
R R
C D
dt
| bσ (c) − bσn (c) | 2 + R σ(t)σ n (t)
≤ ε 1 + R σ(t)σn (t)
C D
dt C D
dt
" #
E 2 C D1 dt
R
ε
| bσ (c) − bσn (c) | ≤ 1 + R σ(t)σn (t)
2 dt
C D

The final step is to establish bounds for D. It is clear that (σn (t)σ(t))2 ≤
D ≤ (E + σn (t))2 (E + σ(t))2 . Using these bounds, we arrive at

E 2 C (E+σn (t))12 (E+σ(t))2 dt

" R #
ε
| bσ (c) − bσn (c) | ≤ 1+ R σ(t)σn (t)
2 dt
C (σn (t)σ(t))2

E 2 C (E+σn (t))12 (E+σ(t))2 dt

" R #
ε
≤ 1+ R 1 .
2 C σn (t)σ(t)
dt

The right hand side of this inequality does not depend on c. This demon-
strates that bσn converges uniformly to bσ . We conclude that the mapping T
is continuous. All of the conditions of the Schauder Fixed Point Theorem are
satisfied, so T has a fixed point. This proves existence of an equilibrium.

16
References
Bernardo, Antonio, Eric Talley, and Ivo Welch 2000. “A Theory of Legal
Presumptions,” Journal of Law, Economics, and Organization, 16, 1–49.

Clark, Derek J. and Christian Riis 1998. “Contest success functions: an

extension,” Economic Theory, 11, 201–204.

Corchón, Luis 2007. “The Theory of Contests: A Survey,” Review of Eco-

nomic Design. forthcoming.

Cornes, Richard and Roger Hartley 2005. “Asymmetric Contests with Gen-
eral Technologies,” Economic Theory, 26, 923–946.

Gradstein, Mark 1995. “Intensity of Competition, Entry and Entry Deter-

rence in Rent Seeking Contests,” Economics and Politics, 7, 79–91.

Harstad, Ronald M. 1995. “Privately Informed Seekers of an Uncertain

Rent,” Public Choice, 83, 81–93.

Higgins, Richard S., William F. Shughart, and Robert D. Tollison 1985. “Free
Entry and Efficient Rent Seeking,” Public Choice, 46, 247–258.

Hillman, Arye L. and Eliakim Katz 1984. “Risk-Averse Rent Seekers and the
Social Cost of Monopoly Power,” The Economic Journal, 94, 104–110.

Hillman, Arye L. and John G. Riley 1989. “Politically contestable rents and
transfers,” Economics and Politics, 1, 17–39.

Hurley, Terrance M. and Jason F. Shogren 1998a. “Asymmetric information

contests,” European Journal of Political Economy, 14, 645–665.

Hurley, Terrance M. and Jason F. Shogren 1998b. “Effort levels in a Cournot

Nash contest with asymmetric information,” Journal of Public Economics,
69, 195–210.

Lockard, Alan A. and Gordon Tullock 2001. Efficient Rent-Seeking: Chroni-

cle of an Intellectual Quagmire. New York: Springer-Verlag.

Malueg, David A. and Andrew J. Yates 2004. “Rent Seeking With Private
Values,” Public Choice, 119, 161–178.

17
Malueg, David A. and Andrew J. Yates 2006. “Equilibria in Rent-Seeking
Contests with Homogeneous Success Functions,” Economic Theory, 27,
719–727.

Nitzan, Shmuel 1991. “Collective Rent Dissipation,” The Economic Journal,

101, 1522–1534.

Schoonbeek, Lambert and Barbara M. Winkel 2006. “Activity and Inactivity

in a Rent-Seeking Contest with Private Information,” Public Choice, 127,
123–132.

Skaperdas, Stergios 1996. “Contest success functions,” Economic Theory, 7,

283–290.

Szidarovsky, Ferenc and Koji Okuguchi 1997. “On the Existence and Unique-
ness of Pure Nash Equilibrium in Rent-Seeking Games,” Games and Eco-
nomic Behavior, 18, 135–140.

Tullock, Gordon 1980. “Efficient Rent Seeking.” In Buchanan, James, Robert

Tollison, and Gordon Tullock, eds., Toward a Theory of the Rent-Seeking
Society, pages 97–112. College Station, TX: Texas A & M Univ. Press.

Van Long, Ngo and Neil Vousden 1987. “Risk-Averse Rent Seeking with
Shared Rents,” The Economic Journal, 97, 971–985.

Wärneryd, Karl 2003. “Information in Conflicts,” Journal of Economic The-

ory, 110, 121–136.

Wärneryd, Karl 2006. “Participation in Contests with Asymmetric Informa-

tion,” Stockholm School of Economics.

Zeidler, Eberhard 1986. Nonlinear functional analysis and its applications. I

Fixed-point theorems. New York: Springer-Verlag. Translated from the
German by Peter R. Wadsack.

18