Equilibrium contingent compensation in contests with delegation.
Baik, Kyung Hwan
1. Introduction
Consider a lawsuit between a plaintiff and a defendant. Each
litigant first hires an attorney and writes a contract with him. Then,
each attorney expends his effort to win the lawsuit on behalf of his
client. Because the outcome of the lawsuit depends on the
attorney's effort, which in turn depends on the contract, the
litigant must take into account the strategic aspects of contracts when
designing her contract.
The purpose of this paper is to consider contests with delegation,
like the illustrative example above, focusing on equilibrium contracts.
(1,2) Specifically, we consider contests in which two players each want
to win a prize, and each player hires a delegate who expends his effort
to win the prize on the player's behalf. We endogenize delegation
contracts between the players and their delegates while explicitly
taking into account the delegates' participation constraints based
on their reservation wages.
Contests with delegation abound. Examples include litigation in
which litigants hire lawyers to win lawsuits; rent-seeking contests in
which firms, organizations, or individuals hire lobbyists to acquire
government favors and business; and research and development contests in
which firms hire research groups or university professors to obtain
patents.
We consider two-player contests with bilateral delegation. The
players are risk-neutral, and Player 1 values the prize more highly than
Player 2. The players design and provide compensation schemes for their
delegates. The delegates are risk-neutral. They have the same
nonnegative reservation wage, and have equal ability for the contest.
The delegates' effort is not verifiable to a third party, which
implies that contracts contingent on the delegates' effort are
precluded. We assume that each delegate's compensation is
contingent on the outcome of the contest--it depends on whether he wins
or loses the prize.
We formally consider the following two-stage game. In the first
stage, each player hires a delegate and writes a contract with him. The
contract specifies how much the delegate will be paid if he wins the
prize and how much if he loses it. Then the players simultaneously
announce the contracts written independently. In the second stage, after
knowing both contracts, the delegates choose their effort levels
simultaneously and independently. At the end of the second stage, the
winner is determined and each player pays compensation to her delegate
according to the contract written in the first stage.
Fershtman and Kalai (1997) distinguish between two types of
delegation: incentive delegation and instructive delegation. In the case
of incentive delegation, a player provides an incentive scheme for her
delegate, and the delegate chooses an effort level that maximizes his
own payoff, given the incentive scheme. In the case of instructive
delegation, a player designs a set of instructions and requires her
delegate to follow the instructions. According to this classification,
then, this paper adopts incentive delegation. The players in this paper
provide compensation schemes for their delegates that are based on the
observables, and the delegates choose their effort levels given the
compensation schemes.
Solving for the subgame-perfect equilibrium of the two-stage game,
we obtain the equilibrium contracts between the players and their
delegates, and show that each player's equilibrium contract is a
no-win-no-pay contract--a contract that specifies zero compensation for
a delegate if he loses the prize. Then, we examine the delegates'
equilibrium compensation spreads, effort levels, probabilities of
winning, expected payoffs, and the players' equilibrium expected
payoffs. We define a delegate's compensation spread as the
difference between what he earns if he wins the prize and what he earns
if he loses it.
We obtain the result of no-win-no-pay contracts because of the
constraint that a delegate's compensation should not be negative if
he loses the prize, and the assumption that the delegates are
risk-neutral. The result of no-win-no-pay contracts makes intuitive
sense. By choosing such a contract, each player makes her
delegate's compensation spread as wide as possible so that she can
most strongly motivate her delegate to win the prize.
Another interesting result is that when a delegate's
participation constraint is not binding in equilibrium, his equilibrium
expected payoff is greater than his reservation wage. Recall that
economic rent is defined as that part of the compensation received by
the owner of a resource that exceeds the resource's opportunity
cost. Then we may say that the gap between the delegate's
equilibrium expected payoff and his reservation wage constitutes the
economic rent for the delegate. This economic rent is not created
because of restrictions on entry into the "delegate industry,"
but created because of both the inability to write contracts based on a
delegate's effort and the players' strategic decisions on
their delegates' compensation. Indeed, competition among potential
delegates to become this particular delegate, if any, cannot reduce the
delegate's equilibrium expected payoff to his reservation wage.
We also obtain: (i) Delegate 1's compensation spread is
greater than Delegate 2's, and (ii) the equilibrium expected payoff
for Delegate 1 is greater than that for Delegate 2. These occur unless
both delegates' participation constraints are binding in the
subgame-perfect equilibrium. Part (i) implies that the player with a
higher valuation--the hungrier player--offers her delegate better
contingent compensation than her opponent does. Part (ii) is very
interesting because the delegates are identical before signing up for
their players: They have equal ability for the contest and have the same
reservation wage. The difference in the delegates' expected payoffs
arises because of the inability to write contracts based on the
delegates' effort and because Player 1 motivates her delegate more
strongly than Player 2--that is, Delegate l's compensation spread
is greater than Delegate 2's. In this case, even though there
exists competition among potential delegates to be employed by Player 1,
it cannot lead to the same expected payoff for the delegates.
The assumption that the delegates' effort is not verifiable to
a third party--which implies the inability to write contracts based on
the delegates' effort--is crucial in obtaining the result that the
economic rents for the delegates exist. Indeed, the economic rent for
each delegate exists because the delegate's effort is his private
information. In this respect, the economic rent for each delegate can be
interpreted as an informational rent, which is a well-known concept in
the principal-agent literature. (3)
There are two main motives of delegation. The first is that a
player wants to use superior ability by hiring a delegate who has more
ability than herself; the second is that a player wants to achieve
strategic commitments through delegation. Baik and Kim (1997) first
introduced delegation into the literature on the theory of contests.
They present a model that involves both motives of delegation.
Considering two-player contests in which each player has the option of
hiring a delegate, they first establish that buying superior ability is
an important motive of delegation. They then show that, as compared with
the model without delegation, a total effort level is less when
unilateral delegation by the player with a higher valuation or bilateral
delegation arises, but it is greater when unilateral delegation by the
player with a lower valuation arises. However, they assume that the
delegation contracts are exogenously given, and assume implicitly that
each delegate's reservation wage is zero. Warneryd (2000) considers
two-player contests with bilateral delegation. He shows that compulsory delegation with moral hazard--that is, where the delegates' effort
is unobservable--may be beneficial to the players. He also shows that
this result holds even when secret renegotiation opportunities are given
to the players and delegates. Schoonbeek (2002) considers a two-player
contest in which only one player, say Player 1, has the option of hiring
a delegate. He compares the equilibrium expected utility of Player 1 in
the unilateral-delegation case with that in the no-delegation case,
focusing on the impact of the risk aversion of Player 1 with respect to
her money income. Konrad, Peters, and Warneryd (2004) consider a
first-price all-pay auction with two buyers in which each buyer has the
option of hiring an agent. They show that in equilibrium each buyer
delegates the bidding to her agent; and both buyers are better off. They
also show that the buyers provide their agents with incentives to make
bids that differ from the bids the buyers would like to make, and the
delegation contracts are asymmetric even if the buyers and the auction
are perfectly symmetric.
The paper proceeds as follows. In section 2, we develop the model
and set up the two-stage game. We then obtain a unique Nash equilibrium of a second-stage subgame. In section 3, we analyze the first stage of
the two-stage game. We first show that each player writes a
no-win-no-pay contract with her delegate. Then we obtain the equilibrium
contracts chosen by the players. Section 4 examines the delegates'
equilibrium compensation spreads, effort levels, probabilities of
winning, their equilibrium expected payoffs, and the players'
equilibrium expected payoffs. Finally, section 5 offers our conclusions.
2. The Model
Consider a contest in which two risk-neutral players, 1 and 2, each
want to win a single indivisible prize, and each player hires a delegate
who expends his effort to win the prize on the player's behalf.
Each delegate's effort may be observable to his employer, but is
not verifiable to a third party. This implies that contracts contingent
on a delegate's effort are precluded. The players' valuations
for the prize differ. Let [v.sub.i] represent Player i's valuation
for the prize. We assume that Player 1 values the prize more highly than
Player 2: [v.sub.1] > [v.sub.2]. Each player's valuation for the
prize is positive and publicly known.
The players design compensation schemes for their delegates: Player
i sets compensation for her delegate, denoted by [W.sub.i] and
[L.sub.i]. Compensation of [W.sub.i] is paid to Delegate i if he wins
the prize, and [L.sub.i] if he loses it. Note that Delegate i's
compensation is contingent on the outcome of the contest. Let [W.sub.i]
= [[alpha].sub.i][v.sub.i] and let [L.sub.i] = [[beta].sub.i][v.sub.i],
where [[beta].sub.i] < [[alpha].sub.i] < 1 and [[beta].sub.i]
[greater than or equal to] 0. (4) Then, since vi is exogenously given,
Player i designs the compensation scheme for her delegate by choosing
the value of [[alpha].sub.i] and that of [[beta].sub.i].
The delegates are risk-neutral. Delegate i has a reservation wage
of [R.sub.i], where [R.sub.i] is nonnegative and is much less than
[v.sub.i]. (5) This implies that when Delegate i signs up for Player i,
his expected payoff must be greater than or equal to his reservation
wage, given the compensation scheme designed by Player i. Otherwise--if
his expected payoff falls short of his reservation wage--Delegate i
prefers not to work for Player i and accepts alternative employment
instead.
We formally consider the following two-stage game. In the first
stage, each player hires a delegate and writes a contract with him--in
other words, Player i designs and offers Delegate i a compensation
scheme, which Delegate i accepts. The contract specifies how much the
delegate will be paid if he wins the prize and how much if he loses it.
Then the players simultaneously announce the contracts written
independently--that is, Player 1 announces publicly the values of
[[alpha].sub.1] and [[beta].sub.1], and Player 2 announces publicly the
values of [[alpha].sub.2] and [[beta].sub.2]. In the second stage, after
knowing both contracts, the delegates choose their effort levels
simultaneously and independently. At the end of the second stage, the
winner is determined and each player pays compensation to her delegate
according to the contract written in the first stage.
In the second stage of the game, the delegates compete with each
other by expending irreversible effort to win the prize. Let [x.sub.i]
represent the effort level expended by Delegate i. Effort levels are
nonnegative and are measured in units commensurate with the prize. Let
[p.sub.1]([x.sub.1], [x.sub.2]) denote the probability that Delegate 1
wins the prize when the delegates' effort levels are [x.sub.1] and
[x.sub.2]. The contest success function for Delegate 1 is given by: (6)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (1)
Let [[pi].sub.i] represent the expected payoff for Delegate i. Then
the payoff function for Delegate 1 is
[[pi].sub.1] = [[beta].sub.1][v.sub.1] + ([[apha].sub.1] -
[[beta].sub.1])[v.sub.1][p.sub.1]([x.sub.1], [x.sub.2]) - [x.sub.1]. (2)
Similarly, the payoff function for Delegate 2 is
[[pi].sub.2] = [[alpha].sub.2][v.sub.2] - ([[alpha].sub.2] -
[[beta].sub.2])[v.sub.2][p.sub.1]([x.sub.1], [x.sub.2]) - [x.sub.2]. (3)
Next, consider the players' expected payoffs computed in the
first stage of the game--when Player i believes that Delegate 1 will
expend an effort level of [x.sub.1] and Delegate 2 will expend an effort
level of [x.sub.2] in the second stage. Given Player i's contract,
([W.sub.i], [L.sub.i]), if her delegate wins the prize in the second
stage, Player i's net payoff will be [v.sub.I] - [W.sub.i];
otherwise, Player i will gain nothing, but should pay [L.sub.i] to her
delegate. Let [G.sub.i] represent the expected payoff for Player i. Then
the payoff function for Player 1 is
[G.sub.1] = -[[beta].sub.1][v.sub.1] + (1 - [[alpha].sub.1], +
[[beta].sub.1])[v.sub.1][P.sub.1]([x.sub.1], [X.sub.2]). (4)
Similarly, the payoff function for Player 2 is
[G.sub.2] = (1 - [[alpha].sub.2])[v.sub.2] - (1 - [[alpha].sub.2] +
[[beta].sub.2])[v.sub.2][p.sub.1]([x.sup.1], [x.sub.2]. (5)
Finally, we assume that all of the above is common knowledge among
the players and delegates. We employ subgame-perfect equilibrium as the
solution concept.
To solve for a subgame-perfect equilibrium of the game, we work
backward. We begin by considering the second stage in which, after
knowing the contracts chosen in the first stage, ([[alpha].sub.1],
[[beta].sub.1]) and ([[alpha].sub.2], [[beta].sub.2]), Delegate i seeks
to maximize his expected payoff over his effort level, given the other
delegate's effort level. Given a positive effort level of Delegate
2, the first-order condition for maximizing Delegate 1's expected
payoff, [[pi].sub.1], yields
([[alpha].sub.1] - [[beta].sub.1])[v.sub.1]([partial
derivative][p.sub.1] ([x.sub.1], [x.sub.2])/[partial
derivative][x.sub.1]) = 1. (6)
Given a positive effort level of Delegate 1, the first-order
condition for maximizing Delegate 2's expected payoff,
[[pi].sub.2], yields
- ([[alpha].sub.2] - [[beta].sub.2])[v.sub.2] ([partial
derivative][p.sub.1] ([x.sub.1], [x.sub.2])/[partial
derivative][x.sub.2]) (7)
Conditions 6 and 7 say that, given the other delegate's
positive effort level, if Delegate i's best response--an effort
level that maximizes his expected payoff--is positive, then his marginal
gross payoff--the left-hand side of Condition 6 or 7--must be equal to
his marginal cost, 1, at that effort level. (7) Delegate i's payoff
function is strictly concave in his effort level. Thus the second-order
condition for maximizing [[pi].sub.i] is satisfied, and Delegate
i's best response is unique.
A Nash equilibrium of the second-stage subgame is a pair of effort
levels, one for each delegate, at which each delegate's effort
level is the best response to his opponent's. Thus it satisfies the
delegates' reaction functions--which are derived from Conditions 6
and 7--simultaneously. We obtain a unique Nash equilibrium:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
and
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (8)
Using Equation 8, we obtain [x.sup.N.sub.1]/[x.sup.N.sub.2] =
([[alpha].sub.1] - [[beta].sub.1])[v.sub.1]/ ([[alpha].sub.2] -
[[beta].sub.2][v.sub.2] or, equivalently,
[x.sup.N.sub.1]/([[alpha].sub.1] - [[beta].sub.1])[v.sub.1] =
[x.sup.N.sub.2]/([[alpha].sub.2] - [[beta].sub.2])[v.sub.2]. (8) Note
that ([[alpha].sub.1] - [[beta].sub.1])[v.sub.i] is the difference
between what Delegate i earns if he wins the prize and what he earns if
he loses it. Let us call it Delegate i's compensation spread. (9)
Then the first expression says that, at the Nash equilibrium, the ratio
of the delegates' effort levels is equal to the ratio of their
compensation spreads. We can explain this result as follows: Since each
delegate's probability of winning is a function of the ratio of the
two delegates' effort levels, the equilibrium effort ratio should
be equal to the compensation-spread ratio in order to satisfy the
mutual-best-responses property of Nash equilibrium. In the second
expression, [x.sup.N.sub.i]/([[alpha].sub.i] - [[beta].sub.i])[v.sub.i]
is the proportion of Delegate i's equilibrium effort level--which
is "dissipated" in pursuit of the prize--to his compensation
spread. The second equation says that these proportions are the same
between the delegates. It follows immediately from Equation 8 that the
proportion is less than a quarter.
3. Equilibrium Contingent Compensation
In this section, we analyze the first stage of the two-stage game.
We first show that each player writes a no-win-no-pay contract with her
delegate--that is, Player i chooses a contract with [[beta].sub.i] = 0.
Then we obtain the equilibrium contracts also called the equilibrium
contingent compensation--chosen by the players.
In the first stage, the players choose their contracts
simultaneously and independently. The players have perfect foresight about the second-stage competition--more specifically, the Nash
equilibrium of each second-stage subgame. Let [p.sub.1]([x.sup.N.sub.1],
[x.sup.N.sub.2]) be the probability that Delegate 1 wins the prize at
the Nash equilibrium of the second-stage subgame, given contracts,
([[alpha].sub.1], [[beta].sub.1]) and ([[alpha].sub.2], [[beta].sub.2]).
Then, using Equations 4 and 5, we obtain the players' payoff
functions that take into account the Nash equilibrium of the
second-stage subgame:
[G.sup.N.sub.1] = -[[beta].sub.1][v.sub.1] + (1 - [[alpha].sub.1] +
[[beta].sub.1])[v.sub.1][p.sub.1]([x.sup.N.sub.1], [x.sup.N.sub.2])
and
[G.sup.N.sub.2] : (1 - [[alpha].sub.2])[v.sub.2] - (1 -
[[alpha].sub.2] + [[beta].sub.2])[v.sub.2][p.sub.1]([x.sup.N.sub.1],
[x.sup.N.sub.1]),
where [p.sub.1]([x.sup.N.sub.1], [x.sup.N.sub.1]) =
([[alpha].sub.1] - [[beta].sub.1])[v.sub.1]/{([[alpha].sub.1] -
[[beta].sub.1])[v.sub.1] + ([[alpha].sub.2] - [[beta].sub.2])[v.sub.2]},
which are obtained using Equations 1 and 8.
When choosing a contract for her delegate, each player should
consider her delegate's participation constraint. Having perfect
foresight about the Nash equilibrium of each second-stage subgame, the
players and delegates can compute, in the first stage, the
delegates' expected payoffs. Using Equations 2 and 3, we obtain the
delegates' payoff functions that are associated with the Nash
equilibrium of the second-stage subgame, given contracts,
([[alpha].sub.1], [[beta].sub.1]) and ([[alpha].sub.2], [[beta].sub.2]):
[[pi].sup.N.sub.1] = [[beta].sub.1][v.sub.1] + ([[alpha].sub.1] -
[[beta].sub.1])[v.sub.1][p.sub.1]([x.sup.N.sub.1], [x.sup.N.sub.2]) -
[x.sup.N.sub.1]
and
[[pi].sup.N.sub.2] = [[alpha].sub.2][v.sub.2] + ([[alpha].sub.2] -
[[beta].sub.2])[v.sub.2][p.sub.1]([x.sup.N.sub.1], [x.sup.N.sub.2]) -
[x.sup.N.sub.2].
Delegate i's participation constraint is then
[[pi].sup.N.sub.i] [greater than or equal to] [R.sub.i].
Now Player i faces the following constrained-maximization problem:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
subject to [[pi].sup.N.sub.i] [greater than or equal to] [R.sub.i.
That is, taking the opponent's contract as given, Player i
seeks to maximize her expected payoff over her contract,
([[alpha].sub.i], [[beta].sub.i]), subject to Delegate i's
participation constraint. By doing so, she obtains her best
response--denoted by ([[alpha].sup.b.sub.i], [[beta].sup.b.sub.i])--to
the given contract of her opponent. To solve for each player's best
response in an "informative" way, we will break up the
constrained-maximization problem into two pieces. First, we will look at
the problem of how to maximize each player's expected payoff
without considering her delegate's participation constraint. Then,
we will look at the problem of how to choose each player's best
response while considering her delegate's participation constraint.
We begin by looking at the first step--maximizing each
player's expected payoff without considering her delegate's
participation constraint. We obtain Lemma 1. (10)
LEMMA 1. (a) Given Player j's contract, ([[alpha].sub.j],
[[beta].sub.j]), and given [[alpha].sub.i], Player i's expected
payoff is always decreasing in [[beta].sub.i]: In terms of the symbols,
we have [partial derivative][G.sup.N.sub.i]/[partial
derivative][[beta].sub.i] < 0. (11) (b) Given Player j's
contract, ([[alpha].sub.j], [[beta].sub.j]), and given [[beta].sub.i],
Player i's expected payoff is maximized at [[alpha].sub.i] =
[[beta].sub.i] + [k.sub.i], where
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Part (a) can be explained as follows. As [[beta].sub.i] decreases,
Delegate i's compensation spread increases. A larger compensation
spread in turn gives Delegate i more incentives to win the prize and
makes him exert more effort. (12) A higher effort level of Delegate i
then yields a higher probability that Delegate i wins the prize in
second-stage equilibrium. Therefore, a higher probability of winning and
less compensation in the case of losing lead to an increase in Player
i's expected payoff.
In the [[beta].sub.i][[alpha].sub.i]-space of Figures 1 and 2, the
graph of [[alpha].sub.i] = [[beta].sub.i] + [k.sub.i] is a straight line
with a vertical intercept of [k.sub.i] and a slope of unity. It is easy
to see that ks is positive but less than a half. It follows from Lemma 1
that, given Player j's contract, ([[alpha].sub.j], [[beta].sub.j]),
Player i's expected payoff is maximized when ([alpha].sub.i],
[[beta].sub.i]) = ([k.sub.i], 0).
[FIGURES 1-2 OMITTED]
Next, we look at the second step--the problem of how to choose each
player's best response while considering her delegate's
participation constraint. Consider first Delegate i's participation
constraint whose weak-inequality sign is replaced by the equals
sign--that is, consider [[pi].sup.N.sub.i] = [R.sub.i]. Figures 1 and 2
illustrate its graph. Let us call it Delegate i's participation
constraint curve. It is straightforward to see that, in the
[[beta].sub.i][[alpha].sub.i]-space, Delegate i's participation
constraint curve slopes downward from left to right, and has a vertical
intercept of [m.sub.i], where [m.sub.i] satisfies
[m.sup.3.sub.i][v.sub.3.sub.i] = [R.sub.i][{[m.sub.i][v.sub.1] +
([[alpha].sub.j] - [[beta].sub.j])[v.sub.j]}.sup.2]. (9)
Player i's contracts that satisfy her delegate's
participation constraint, [[pi].sup.N.sub.i] [greater than or equal to]
[R.sub.i], lie on or above her delegate's participation constraint
curve. Thus they are located in the shaded areas of Figures 1 and 2.
Given Player j's contract, ([[alpha].sub.j], [[beta].sub.j]),
because [m.sub.i] increases in [R.sub.i] while [k.sub.i] is independent
of [R.sub.i], we have two different cases depending on the size of
Delegate i's reservation wage, [R.sub.i]. Figure 1 shows the first
case where Delegate i's participation constraint is not binding:
[k.sub.i] > [m.sub.i]. This case occurs when [R.sub.i] is
"low." Figure 2 shows the second case where Delegate i's
participation constraint is binding: [k.sub.i] [less than or equal to]
[m.sub.i]. This case occurs when [R.sub.i] is "high."
Given Player j's contract, ([[alpha].sub.j], [[beta].sub.j]),
Player i's best response to ([[alpha].sub.j], [[beta].sub.j]) is
defined as a contract that maximizes her expected payoff,
[G.sup.N.sub.i], subject to Delegate i's participation constraint,
[[pi].sup.N.sub.i] [greater than or equal to] [R.sub.i]. Denote it by
([[alpha].sup.b.sub.i], [[beta].sup.b.sub.i]). Using Lemma 1, we obtain
Lemma 2.
LEMMA 2. (a) In the case where [R.sub.i] is low, Delegate i's
participation constraint is not binding, and Player i's best
response to Player j's contract, ([[alpha].sub.j], [[beta].sub.j]),
is ([k.sub.i], 0): In terms of the symbols, we have [k.sub.i] >
[m.sub.i] and ([[alpha].sup.b.sub.i], [[beta].sup.b.sub.i]) =
([k.sub.i], 0) when [R.sub.i] is low. (b) In the case where [R.sub.i] is
high, Delegate i's participation constraint is binding, and Player
i's best response is ([m.sub.i], 0): In terms of the symbols, we
have [k.sub.i] [less than or equal to] [m.sub.i] and
([[alpha].sup.b.sub.i], [[beta].sup.b.sub.i]) = ([m.sub.i], 0) when
[R.sub.i] is high.
Part (a) says that, when Delegate i's reservation wage is low,
Player i chooses a contract that gives Delegate i an expected payoff
higher than his reservation wage. The explanation for this follows.
Delegate i will compete against Delegate j to win the prize in the
second stage. Player i wants to induce Delegate i to exert the
"optimal" effort--the optimal effort for Player i--by choosing
the "best" contract, given Player j's contract,
([[alpha].sub.j], [[beta].sub.j]). In this case, the "best"
contract--that maximizes Player i's expected payoff when her
delegate's participation constraint is absent--happens to yield
Delegate i's expected payoff greater than his reservation wage,
because his reservation wage is low. While Player i looks benevolent,
she is actually pursuing her self-interest.
In the case where Delegate i's reservation wage is high, the
"best" contract--the solution to the
unconstrained-maximization problem--yields Delegate i's expected
payoff less than his reservation wage. Hence, to take care of her
delegate's participation constraint, Player i chooses a contract
that lies on Delegate i's participation constraint curve.
Now, we obtain the equilibrium contracts chosen by the players. Let
([[alpha].sup.*.sub.i], [[beta].sup.*.sub.i]) represent Player i's
contract that is specified in the subgame-perfect equilibrium of the
two-stage game. We first obtain from Lemma 2 that [[beta].sup.*.sub.1] =
[[beta].sup.*.sub.2] = 0. In order to obtain [[alpha].sup.*.sub.1] and
[[alpha].sup.*.sub.2], we utilize the players' reaction curves in
the [[alpha].sub.1][[alpha].sub.2]-space. It follows immediately from
Lemma 2 that, given [[alpha].sup.*.sub.j] = 0, Player i's reaction
curve in the [[alpha].sub.1][[alpha].sub.2]-space is the graph of
[[alpha].sup.b.sub.i] = max{[k.sup.o.sub.i], [m.sup.o.sub.i]), where
[k.sup.o.sub.i] = {-[[alpha].sub.j][v.sub.j] +
[([[alpha].sup.2.sub.j][v.sup.2.sub.j] +
[[alpha].sub.j][v.sub.j][v.sub.j]).sup.1/2]}/[v.sub.i] and
[m.sup.o.sub.i] satisfies [([m.sup.o.sub.i][v.sub.i]).sup.3] =
[R.sub.i][([m.sup.o.sub.i][v.sub.i] + [[alpha].sub.j][v.sub.j]).sup.2],
which are based on Lemma 1 and Equation 9, respectively. Then, the
intersection of the two reaction curves determines [[alpha].sup.*.sub.1]
and [[alpha].sup.*.sub.2] Because [m.sup.o.sub.i]--depends on--more
specifically, increases in--Delegate i's reservation wage,
[R.sub.i], the equilibrium contracts of the players depend on the
delegates' reservation wages. Henceforth, to get more mileage, we
assume that the delegates have the same reservation wage: [R.sub.1] =
[R.sub.2] = R.
Figure 3 is useful in obtaining [[alpha].sup.*.sub.1] and
[[alpha].sup.*.sub.2]. For concise exposition, we draw the graphs of
[k.sup.o.sub.i] and [m.sup.o.sub.i] separately rather than draw the
graph of [[alpha].sup.b.sub.1] = max{[k.sup.o.sub.i], [m.sup.o.sub.i]},
which is Player i's reaction curve. Lemma A1 in the Appendix
describes properties of the graphs in Figure 3. Lemma 3 describes the
equilibrium contracts of the players, ([a.sup.*.sub.1],
[[beta].sup.*.sub.1]) and ([[alpha].sub.1] [[beta].sup.*.sub.2]).
[FIGURE 3 OMITTED]
LEMMA 3. (a) If the intersection of the graphs of [m.sup.o.sub.1]
and [m.sup.o.sub.2] lies on line segment OA, or equivalently, if 0 [less
than or equal to] R < [R.sup.A], then ([[alpha].sup.*.sub.1],
[[alpha].sup.*.sub.2]) occurs at point Q--the intersection of the graphs
of [k.sup.o.sub.1] and [k.sup.o.sub.2]. (b) If the intersection of the
graphs of [m.sup.o.sub.1] and [m.sup.o.sub.2] lies on line segment AD,
or equivalently, if [R.sup.A] [less than or equal to] R < [R.sup.D],
then ([[alpha].sup.*.sub.1], [[alpha].sup.*.sub.2]) occurs at the
intersection, on are QD, of the graphs of [k.sub.o.sub.1] and
[m.sup.o.sub.2]. (c) If the intersection of the graphs of
[m.sup.o.sub.1] and [m.sup.o.sub.2] lies on line segment DS, or
equivalently, if [R.sup.D] [less than or equal to] R < [v.sub.2]/4,
then ([[alpha].sup.*.sub.1], [[alpha].sup.*.sub.2]) occurs at this very
intersection: ([[alpha].sup.*.sub.1], [[alpha].sup.*.sub.2]) =
(4R/[v.sub.1], 4R/[v.sub.2]). (13) (d) We obtain [[beta].sup.*.sub.1] =
[[beta].sup.*.sub.2] = 0, regardless of the value of R, where 0 [less
than or equal to] R < [v.sup.2]/4.
Lemma 3 says that the equilibrium contracts of the players are
no-win-no-pay contracts. More specifically, the equilibrium contract of
Player i specifies that Delegate i earns [W.sup.*.sub.i] =
[[alpha].sup.*.sub.i][v.sub.i] if he wins the prize, and [L.sup.*.sub.i]
= [[beta].sup.*.sub.i][v.sub.i] = 0 if he loses it. (14) This means that
Delegate i's compensation spread in the subgame-perfect equilibrium
is [[alpha].sup.*.sub.i][v.sub.i]. Why does each player choose a
no-win-no-pay contract? A convincing reason is that, by doing so, each
player can most strongly motivate her delegate to win the prize. Indeed,
by choosing such a contract, each player makes her delegate's
compensation spread--the gap between what her delegate earns if he wins
the prize and what he earns if he loses it--as wide as possible. Then,
facing such a contract, Delegate i tries his best to win the prize in
the second stage, which is beneficial to Player i.
Lemma 3 implies that, as the delegates' reservation wage
increases beyond [R.sup.A], the delegates' equilibrium compensation
spreads--or their equilibrium contingent fees--increase. This can be
explained as follows. First, when the reservation wage increases, the
players must offer their delegates higher compensation spreads in order
to hire them. Second, when the opponent offers a higher compensation
spread to her delegate, each player has an incentive to follow suit.
Facing a more aggressive delegate of the opponent, each player must make
her delegate more aggressive by increasing his compensation spread.
Lemma 3 establishes that there are three possible types of the
equilibrium-contracts pairs: the pairs of contracts at which neither of
the delegates' participation constraints is binding; the pairs of
contracts at which Delegate 2's participation constraint is
binding, but Delegate 1's is not; and the pairs of contracts at
which both delegates' participation constraints are binding. The
first type, called Type I, is associated with Part (a) of Lemma 3; the
second type, called Type II, is associated with Part (b); and the third
type, called Type III, is associated with Part (c).
4. Three Types of Equilibrium-Contracts Pairs
In this section, we closely look at the three types of
equilibrium-contracts pairs, and examine the delegates'
compensation spreads, their effort levels, their probabilities of
winning, their expected payoffs, and the players' expected payoffs.
Let ([[alpha].sup.*.sub.1], [[alpha].sup.*.sub.2]) represent the
effort levels of the delegates that are specified in the subgame-perfect
equilibrium. Let [p.sub.1] ([[alpha].sup.*.sub.1],
[[alpha].sup.*.sub.2]) be the probability that Delegate 1 and thus
Player 1 win the prize in the subgame-perfect equilibrium. Let
[[pi].sup.*.sub.i] and [G.sup.*.sub.i] represent the expected payoff for
Delegate i and Player i, respectively, in the subgame-perfect
equilibrium. Then, using Lemma 3 and Equations 1-5 and 8, we obtain
Proposition 1. (15)
PROPOSITION 1. [Type I] In the case where the delegates'
reservation wages are low, and thus neither of the delegates'
participation constraints is binding in equilibrium, we obtain: (a)
[[alpha].sup.*.sub.1][v.sub.1] > [[alpha].sup.*.sub.2][v.sub.2] and
[[alpha].sup.*.sub.1] < [[alpha].sup.*.sub.2], (b)
[[alpha].sup.*.sub.1] > [[alpha].sup.*.sub.2], (c)
[p.sub.1]([x.sup.*.sub.1], [x.sup.*.sub.2]) > 1/2, (d)
[[pi].sup.*.sub.1] > [[pi].sup.*.sub.2] > R, and (e)
[G.sup.*.sub.1] > [G.sup.*.sub.2] > 0. [Type II] In the case where
the delegates' reservation wage is rather high, and thus Delegate
2's participation constraint is binding but Delegate 1's is
not in equilibrium, we obtain: [[alpha].sup.*.sub.1][v.sub.1] >
[[alpha].sup.*.sub.2][v.sub.2] and [[alpha].sup.*.sub.1] <
[[alpha].sup.*.sub.2], (b) [[alpha].sup.*.sub.1] >
[[alpha].sup.*.sub.2], (c) [p.sub.1]([x.sup.*.sub.1], [x.sup.*.sub.2])
> 1/2, (d) [[pi].sup.*.sub.1] > [[pi].sup.*.sub.2] = R, and (e)
[G.sup.*.sub.1] > [G.sup.*.sub.2] > 0. [Type III] In the case
where the delegates' reservation wage is high, and thus both
delegates' participation constraints are binding in equilibrium, we
obtain: [[alpha].sup.*.sub.1][v.sub.1] = [[alpha].sup.*.sub.2][v.sub.2]
and [[alpha].sup.*.sub.1] < [[alpha].sup.*.sub.2], (b)
[[alpha].sup.*.sub.1] = [[alpha].sup.*.sub.2], (c)
[p.sub.1]([x.sup.*.sub.1], [x.sup.*.sub.2]) = 1/2, (d)
[[pi].sup.*.sub.1] = [[pi].sup.*.sub.2] = R, and (e) [G.sup.*.sub.1]
> [G.sup.*.sub.2] > 0
Proposition 1 is summarized in Table 1. Consider Type I of the
equilibrium-contracts pairs. First of all, note that the results hold
true even though the delegates' reservation wages differ, as far as
their participation constraints are not binding in the subgame-perfect
equilibrium. Part (a) says that Delegate 1's compensation spread is
greater than Delegate 2's. In other words, the player with a higher
valuation--the hungrier player--offers her delegate better contingent
compensation than her opponent does. Confronting Player 1--the hungrier
player--Player 2 tries to overcome her relative "weakness" in
the valuations for the prize by choosing [[alpha].sup.*.sub.2] which is
greater than [[alpha].sup.*.sub.1]. However, this turns out to be not
enough to make her delegate more aggressive than her opponent's.
Indeed, Delegate 1 exerts more effort than Delegate 2, and thus becomes
the favorite because Delegate 1's compensation spread--or his
contingent fee--is greater than Delegate 2's. (16)
Another interesting result is that the equilibrium expected payoff
for Delegate 1 is greater than that for Delegate 2. This result is very
interesting because the delegates are identical before signing up for
their players: They have the same reservation wage and have equal
ability for the contest (see Eqn. 1). (17) The result arises due to the
inability to write contracts based on the delegates' effort and
because Player 1 motivates her delegate more strongly than Player
2--that is, Delegate 1's compensation spread is greater than
Delegate 2's. In this case, even though there exists competition
among potential delegates to be employed by Player 1, it cannot lead to
the same expected payoff for the delegates. Delegate 2 signs up for
Player 2 because his equilibrium expected payoff is greater than his
reservation wage. But he would be luckier if he were selected as
Delegate 1 by Player 1's "random drawing."
Yet another interesting result is that each delegate's
equilibrium expected payoff is greater than his reservation wage. The
gap between his equilibrium expected payoff and his reservation wage
constitutes the economic rent for the delegate. This economic rent for
each delegate is not created due to restrictions on entry into the
"delegate industry," but created due to both the inability to
write contracts based on a delegate's effort and the players'
strategic decisions on their delegates' compensation. (18)
Part (e) says that the equilibrium expected payoff for Player 1 is
greater than that for Player 2. This follows immediately from the
assumption that [v.sub.1] > [v.sub.2] and from the results that
[[beta].sup.*.sub.1] = [[beta].sup.*.sub.2] = 0, [[alpha].sup.*.sub.1]
< [[alpha].sup.*.sub.1], and [p.sub.1]([[alpha].sup.*.sub.1],
[[alpha].sup.*.sub.2]) > 1/2, and makes intuitive sense.
Next, consider Type II of the equilibrium-contracts pairs. We
obtain the same results as for Type I with the exception that Delegate
2's equilibrium expected payoff is equal to his reservation wage.
For Type II of the equilibrium-contracts pairs, given Player 1's
equilibrium contract, Player 2's "best" contract--that
maximizes Player 2's expected payoff when her delegate's
participation constraint is absent--yields Delegate 2's expected
payoff less than his reservation wage, because his reservation wage is
rather high. Hence, to hire a delegate, Player 2 must offer her delegate
better contingent compensation that guarantees the delegate his
reservation wage. Note that, as the delegates' reservation wage
increases, the gap between the delegates' equilibrium expected
payoffs narrows. This is because Delegate 2's equilibrium expected
payoff increases--since his participation constraint is binding--as R
increases, while Delegate 1's is "independent" of the
reservation wage. Since [v.sub.1] > [v.sub.2]--more precisely,
[[alpha].sup.*.sub.1][v.sub.1] > [[alpha].sup.*.sub.2][v.sub.2]--and
the delegates have the same reservation wage, Delegate 1's
participation constraint is not binding. Player 1 still has some
breathing space.
Finally, consider Type III: the pairs of contracts at which both
delegates' participation constraints are binding. Part (a) says
that Delegate 1's compensation spread is equal to Delegate
2's. Since both delegates' participation constraints are
binding in this case, the delegates with the same reservation wage must
be treated equally in terms of their compensation--that is, their
equilibrium expected payoffs must be the same and equal to their common
reservation wage. Hence, given [[beta].sup.*.sub.1] =
[[beta].sup.*.sub.2] = 0, we must have [[alpha].sup.*.sub.1][v.sub.1] =
[[alpha].sup.*.sub.2][v.sub.2]. Part (a) also says that
[[alpha].sup.*.sub.2] is greater than [[alpha].sup.*.sub.1]. (19) This
follows immediately from [[alpha].sup.*.sub.1][v.sub.1] =
[[alpha].sup.*.sub.2][v.sub.2] and [v.sub.1] > [v.sub.2].
Parts (b), (c), and (d) show that the delegates expend the same
effort level, have the same probability of winning, and have the same
expected payoff in equilibrium. This is natural because the delegates
are motivated equally to win the prize with the same compensation
spread.
Part (e) says that the equilibrium expected payoff for Player 1 is
greater than that for Player 2. This follows immediately from the
assumption that [v.sub.1] > [v.sub.2] and from the results that
[[beta].sup.*.sub.1] = [[beta].sup.*.sub.2] = 0 and
[[alpha].sup.*.sub.1] < [[alpha].sup.*.sub.2]. The players offer
their delegates the same compensation spread, so that the delegates
exert the same effort and therefore end up with the even contest--that
is, both delegates have the same probability of winning in equilibrium.
However, Player 1's expected payoff is greater than Player 2's
because Player 1 values the prize more highly than Player 2.
5. Conclusions
We have considered contests in which two players each want to win a
prize, and each player hires a delegate who expends his effort to win
the prize on the player's behalf. After obtaining the equilibrium
contracts between the players and their delegates, we have examined the
delegates' equilibrium compensation spreads, effort levels,
probabilities of winning, expected payoffs, and the players'
equilibrium expected payoffs.
First we have shown that each player chooses a no-win-no-pay
contract in equilibrium--that is, Player i chooses a contract with
[[beta].sup.*.sub.1] = 0--and explained that she does so in order to
most strongly motivate her risk-neutral delegate to win the prize.
Then we have found that there are three types of the
equilibrium-contracts pairs, depending on the size of the
delegates' reservation wage: the pairs of contracts at which
neither of the delegates' participation constraints is binding; the
pairs of contracts at which Delegate 2's participation constraint
is binding but Delegate 1's is not; and the pairs of contracts at
which both delegates' participation constraints are binding. For
the first two types of equilibrium-contracts pairs, Delegate 1's
equilibrium compensation spread is greater than Delegate 2's;
Delegate 1's equilibrium effort level is greater than Delegate
2's; and Delegate 1's equilibrium expected payoff is greater
than Delegate 2's. For the third type of equilibrium-contracts
pairs, Delegate 1's equilibrium compensation spread, effort level,
and his equilibrium expected payoff are equal to Delegate 2's,
respectively. For all three types of equilibrium-contracts pairs, the
equilibrium expected payoff for Player 1 is greater than that for Player
2.
We have assumed that Player 1 values the prize more highly than
Player 2. In the case where the players value the prize equally, only
symmetric types of the equilibrium-contracts pairs occur, depending on
the size of the delegates' reservation wage: the pairs of contracts
at which both delegates' participation constraints are not binding,
and the pairs of contracts at which both delegates' participation
constraints are binding. Asymmetric equilibrium-contracts pairs at which
one delegate's participation constraint is binding while the other
delegate's is not, vanish in this symmetric case. For both types of
equilibrium-contracts pairs, the delegates' equilibrium
compensation spreads, effort levels, their equilibrium expected payoffs,
and the players' equilibrium expected payoffs are the same.
We have assumed that [[beta].sub.i] can take on only nonnegative
values. What happens if we set a negative number as the lower bound of
[[beta].sub.i]? First of all, the equilibrium contract of each player is
no longer a no-win-no-pay contract. Instead, it may specify the lower
bound as the equilibrium value of [[beta].sub.i]. This means that
Delegate i is required to pay the absolute value of
[[beta].sub.i][v.sub.i] to his employer if he loses the prize. Or,
following the fixed-fee interpretation in Footnote 4, he is required to
pay the amount to his employer, regardless of the outcome of the
contest. To put it differently, Player i sells Delegate i--for the
amount--both the right to compete for the prize and the right to share
the prize with her when he wins it. (20) Second, if the lower bound of
[[beta].sub.i] is a sufficiently small negative number, there may be no
economic rents for the delegates in equilibrium or the gap between the
delegates' equilibrium expected payoffs.
We have assumed that the delegates' effort is not verifiable
to a third party. If the delegates' effort is observable and
verifiable, so that contracts can be written based on their effort, then
there may be no economic rents for the delegates in equilibrium or the
gap between the delegates' equilibrium expected payoffs. Indeed,
this happens when the players adopt the following compensation structure
for their delegates: A delegate is paid zero if his effort is below a
stipulated effort, and a positive amount if his effort is greater than
or equal to the stipulated effort.
We have assumed that the delegates have equal ability for the
contest, and have the same reservation wage. By doing so, we have set
aside the question of who hires whom. We may be able to endogenize
choice of delegate types, by differentiating delegate types by
delegates' ability for the contest; by introducing delegate types
into contest success functions; and by letting each delegate's
reservation wage depend on his type. This alternative model is an
interesting, natural extension of the paper, but we may have difficulty
in choosing a specific form of the function that describes the
relationship between delegate type and reservation wage.
A further extension of this paper is a model that incorporates
players' decisions on delegation. Baik and Kim (1997) endogenize
players' decisions on delegation. But they assume that the
contracts between the players and their delegates are exogenously given,
and assume implicitly that each delegate's reservation wage is
zero.
Finally, we have assumed that the upper bound of [[alpha].sub.i] is
unity. A model with a lower cap on [[alpha].sub.i] may yield interesting
results. (21) We leave all these considerations for future research.
Appendix
Properties of the Graphs in Figure 3
LEMMA A1. (a) [k.sup.o.sub.i] is increasing in [[alpha].sub.j] at a
decreasing rate. (b) [m.sup.o.sub.i] is increasing in [[alpha].sub.j]at
a decreasing rate. (c) As the delegates' reservation wage, R,
increases, the graph of [m.sup.o.sub.1] shifts to the right while the
graph of [m.sup.o.sub.2] shifts upward. (d) The intersection of the
graphs of [m.sup.o.sub.1] and [m.sup.o.sub.2] always occurs on straight
line OS, which is the graph of [[alpha].sub.2] =
([v.sub.1]/[v.sub.2])[[alpha].sub.1]. (e) Point Q--the intersection of
the graphs of [k.sup.o.sub.1] and [k.sup.o.sub.2]--lies between straight
line OS and the 45[degrees] line. (f) The graph of [k.sup.o.sub.2] cuts
straight line OS at point H. If the graph of [m.sup.o.sub.2] also passes
through point H, then, at the point, the slope of the graph of
[m.sup.o.sub.2] is greater than that of the graph of [k.sup.o.sub.2].
The proof of Lemma A1 is straightforward and therefore omitted.
Lemma A1 says that, as the delegates' reservation wage, R,
increases, the intersection of the graphs of [m.sup.o.sub.1] and
[m.sup.o.sub.2] moves up along straight line OS. However, note that, as
the delegates' reservation wage changes, the graphs of
[k.sup.o.sub.1] and [k.sup.o.sub.2] remain unchanged because
[k.sup.o.sub.i] is independent of the delegates' reservation wage.
In Figure 3, the graph of [k.sup.o.sub.1] cuts straight line OS at point
D and the graph of [k.sup.o.sub.2] cuts straight line OS at point H. It
is easy to see that we have ([[alpha].sub.1], [[alpha].sub.2]) = (1/3,
[v.sub.1]/3[v.sub.2]) at point D and ([[alpha].sub.1],[[alpha].sub.2]) =
([v.sub.2]/3[v.sub.1],1/3) at point H.
Based on Part (f) of Lemma A1, we can draw the graph of
[m.sup.o.sub.2] that passes through both Point A--a point on line
segment OH--and Point Q. Let [R.sup.A] be the value of R that is
associated with this particular graph of [m.sup.o.sub.2], denoted by
[m.sup.o.sub.2] ([R.sup.A]) in Figure 3. Let [R.sup.D] be the value of R
that is associated with the graph of [m.sup.o.sub.2] passing through
point D, denoted by [m.sup.o.sub.2]([R.sup.D]) in Figure 3. We obtain
[R.sup.D] = [v.sub.1]/12. The explicit solution for [R.sup.A] is
unobtainable.
References
Baik, Kyung Hwan. 1994. Effort levels in contests with two
asymmetric players. Southern Economic Journal 61:367-78.
Baik, Kyung Hwan. 2004. Two-player asymmetric contests with
ratio-form contest success functions. Economic Inquiry 42:679-89.
Baik, Kyung Hwan, and In-Gyu Kim. 1997. Delegation in contests.
European Journal of Political Economy 13:281-98.
Baye, Michael R., Dan Kovenock, and Casper G. de Vries. 1993.
Rigging the lobbying process: An application of the all-pay auction.
American Economic Review 83:289-94.
Burtraw, Dallas. 1993. Bargaining with noisy delegation. Rand Journal of Economics 24:40-57.
Che, Yeon-Koo, and Ian L. Gale. 1998. Caps on political lobbying.
American Economic Review 88:643-51.
Clark, Derek J., and Christian Riis. 1998. Competition over more
than one prize. American Economic Review 88:276-89.
Danzon, Patricia Munch. 1983. Contingent fees for personal injury
litigation. Bell Journal of Economics 14:213-24.
Das, Satya P. 1997. Strategic managerial delegation and trade
policy. Journal of International Economics 43:173-88.
Dixit, Avinash. 1987. Strategic behavior in contests. American
Economic Review 77:891-98.
Ellingsen, Tore. 1991. Strategic buyers and the social cost of
monopoly. American Economic Review 81:648-57. Emons, Winand. 2000.
Expertise, contingent fees, and insufficient attorney effort.
International Review of Law and Economics 20:21-33.
Fershtman, Chaim, and Kenneth L. Judd. 1987. Equilibrium incentives
in oligopoly. American Economic Review 77:927-40.
Fershtman, Chaim, Kenneth L. Judd, and Ehud Kalai. 1991. Observable
contracts: Strategic delegation and cooperation. International Economic
Review 32:551-59.
Fershtman, Chaim, and Ehud Kalai. 1997. Unobserved delegation.
International Economic Review 38:763-74.
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. 1998. Effort levels in a
Cournot Nash contest with asymmetric information. Journal of Public
Economics 69:195-210.
Katz, Michael L. 1991. Game-playing agents: Unobservable contracts
as precommitments. Rand Journal of Economics 22:307-28.
Konrad, Kai A. 2000. Trade contests. Journal of International
Economies 51:317-34.
Konrad, Kai A., Wolfgang Peters, and Karl Warneryd. 2004.
Delegation in first-price all-pay auctions. Managerial and Decision
Economics 25:283-90.
Lawarree, Jacques, and Dongsoo Shin. 2005. Organizational
flexibility and cooperative task allocation among agents. Journal of
lnstitutional and Theoretical Economics 161:621-35.
Loury, Glenn C. 1979. Market structure and innovation. Quarterly
Journal of Economies 93:395-410.
Nitzan, Shmuel. 1991. Collective rent dissipation. Economic Journal
101 : 1522-34.
Nitzan, Shmuel. 1994. Modelling rent-seeking contests. European
Journal of Political Economy 10:41-60.
Nti, Kofi O. 1997. Comparative statics for contests and
rent-seeking games. International Economic Review 38:43-59.
Rasmusen, Eric. 2001. Games and information." An introduction
to game theory. 3rd edition. Oxford: Blackwell.
Ray, Tridip. 1999. Share tenancy as strategic delegation. Journal
of Development Economies 58:45-60.
Rosen, Sherwin. 1986. Prizes and incentives in elimination
tournaments. American Economic Review 76:701-15.
Rubinfeld, Daniel L., and Suzanne Scotchmer. 1993. Contingent fees
for attorneys: An economic analysis. Rand Journal of Economics
24:343-56.
Santore, Rudy, and Alan D. Viard. 2001. Legal fee restrictions,
moral hazard, and attorney rents. Journal of Law and Economics
44:549-72.
Schelling, Thomas C. 1960. The strategy of conflict. Cambridge, MA:
Harvard University Press.
Schoonbeek, Lambert. 2002. A delegated agent in a winner-take-all
contest. Applied Economies Letters 9:21-23.
Segendorff, Bjorn. 1998. Delegation and threat in bargaining. Games
and Economic Behavior 23:266-83.
Sklivas, Steven D. 1987. The strategic choice of managerial
incentives. Rand Journal of Economics 18:452-58. Stein, William E., and
Amnon Rapoport. 2004. Asymmetric two-stage group rent-seeking:
Comparison of two contest structures. Public Choice 118:151-67.
Szymanski, Stefan. 2003. The economic design of sporting contests.
Journal of Economic Literature 41:1137-87.
Tullock, Gordon. 1980. Efficient rent seeking. In Toward a theory
of the rent-seeking society, edited by James M. Buchanan, Robert D.
Tollison, and Gordon Tullock. College Station, TX: Texas A&M
University Press, pp. 97-112.
Vickers, John. 1985. Delegation and the theory of the firm.
Economie Journal--Conference Papers Suppl. 95:138-47.
Warneryd, Karl. 2000. In defense of lawyers: Moral hazard as an aid
to cooperation. Games and Economic Behavior 33:145-58.
I am grateful to Yoram Barzel, Rob Gilles, Hans Hailer, Seong Hoon Jeon, Amoz Kats, Fahad Khalil, Chuck Mason, Steve Millsaps, Tim Perri,
Djavad Salehi-Isfahani, Jay Shogren, Karl Warneryd, Kwan Koo Yun, two
anonymous referees, and seminar participants at Appalachian State
University, University of Washington, University of Wyoming, and
Virginia Polytechnic Institute and State University for their helpful
comments and suggestions. Earlier versions of this paper were presented
at the 2003 Annual Meetings of the Allied Social Science Associations,
Washington, D.C., January 2003; and the 2003 Annual Conference of the
Korean Econometric Society, Seoul, Korea, February 2003. Part of this
research was conducted while I was a Visiting Professor at Virginia
Polytechnic Institute and State University. This work was supported by
63 Research Fund, Sungkyunkwan University, 2001.
Received December 2004; accepted May 2006.
Kyung Hwan Baik, Department of Economics, Sungkyunkwan University,
Seoul 110-745, South Korea; E-mail khbaik@skku.edu.
(1) A contest is defined as a situation in which players compete
with one another by expending irreversible effort to win a prize.
Contests have been studied by many economists: Loury (1979), Tullock
(1980), Rosen (1986), Dixit (1987), Hillman and Riley (1989), Ellingsen
(1991), Nitzan (1991, 1994), Baye, Kovenock, and de Vries (1993), Baik
(1994, 2004), Nti (1997), Che and Gale (1998), Clark and Riis (1998),
Hurley and Shogren (1998), Konrad (2000), and Szymanski (2003), to name
a few.
(2) Ever since Schelling (1960) pointed out the benefit of
strategic delegation, many economists have studied delegation in
different contexts. For example, Vickers (1985), Fershtman and Judd
(1987), Sklivas (1987), and Das (1997) have studied strategic managerial
delegation; Burtraw (1993) and Segendorff (1998) have studied delegation
in bargaining situations; Fershtman, Judd, and Kalai (1991), Katz
(1991), and Fershtman and Kalai (1997) have studied delegation with
observable contracts, delegation with unobservable contracts, and
unobserved delegation, respectively; Ray (1999) has studied share
tenancy; Baik and Kim (1997), Warneryd (2000), Schoonbeek (2002), and
Konrad, Peters, and Warneryd (2004) have studied delegation in contests.
(3) For an explanation about the concept of informational rent, see
Rasmusen (2001).
(4) Looking at this compensation structure, one may say that
[[beta].sub.i][v.sub.i] represents a fixed fee that is paid to Delegate
i, regardless of the outcome of the contest, while ([[alpha].sub.1] -
[[beta].sub.i])[v.sub.i] is a contingent fee that is paid to Delegate i
only if he wins the prize. The compensation structure that comprises a
fixed fee and a contingent fee is the standard form of contract between
litigants and attorneys in personal injury and medical malpractice litigation in the United States. See, for example, Danzon (1983),
Rubinfeld and Scotchmer (1993), and Santore and Viard (2001) for
details. The American Bar Association Model Rules of Professional
Conduct require that fixed fees in such tort litigation should not be
negative (see Santore and Viard 2001). This justifies the nonnegativity
constraint on [[beta].sub.i].
(5) However, we will assume in section 3 that the delegates have
the same reservation wage. By doing so, we will make our model more
tractable and will not address the question of who hires whom. This
question is interesting but difficult. We leave it for future research.
We will also assume in section 3 that the delegates' reservation
wage is less than [v.sub.2]/4.
(6) We use simplest logit-form contest success functions throughout
the paper. They are extensively used in the contest literature. See, for
example, Tullock (1980), Hillman and Riley (1989), Ellingsen (1991),
Nitzan (1991), Hurley and Shogren (1998), Warneryd (2000), Szymanski
(2003), and Stein and Rapoport (2004).
(7) Note that, given zero effort of his opponent, each
delegate's best response is to expend infinitesimal effort.
(8) Any "well-behaved" contest success functions that are
homogeneous of degree zero in the delegates' effort levels yield
the same result. Note that, given homogeneous-of-degree-zero contest
success functions, each delegate's probability of winning depends
only on the ratio of the two delegates' effort levels.
(9) Also, as mentioned in Footnote 4, it can be interpreted as a
contingent fee that is paid to Delegate i only if he wins the prize.
(10) To shorten the paper, we omit the proofs of Lemmas 1 and 2.
They are available from the author upon request.
(11) Throughout the paper, when we use i and j at the same time, we
mean that i [not equal to] j.
(12) Indeed, using Equation 8, we can show that, as [[beta].sub.i]
decreases, the equilibrium effort level of Delegate i in the second
stage increases: [partial derivative][x.sup.N.sub.i]/[partial
derivative][[beta].sub.i] < 0. We can also show that, as
[[beta].sub.i] decreases, the probability that Delegate i wins the prize
in second-stage equilibrium increases: [partial derivative][p.sub.1]
([x.sup.N.sub.1], [x.sup.N.sub.2])/[partial derivative][[beta].sub.1]
< 0 and [partial derivative][p.sub.1], ([x.sup.N.sub.1],
[x.sup.N.sub.2])/[partial derivative][[beta].sub.2] > 0.
(13) We must assume that R < [v.sub.2]/4. Otherwise, that is, if
R [greater than or equal to] [v.sub.2]/4, Player 2 ends up with a
nonpositive expected payoff, which implies that Player 2 has no
incentive to hire a delegate at the beginning. This cap on the
reservation wage may seem to exclude interesting cases. However, this is
not so because each delegate's expected payoff is defined as his
gross expected payoff minus his effort level.
(14) Following the interpretation of the compensation structure in
Footnote 4, we can view this equilibrium contingent compensation as
follows: Delegate i's fixed fee, which is paid to him, regardless
of the outcome of the contest, is zero, and his contingent fee, which is
paid to him only if he wins the prize, is [x.sup.*.sub.i][v.sub.i].
(15) To shorten the paper, we omit the proof of Proposition 1. It
is available from the author upon request.
(16) Dixit (1987) calls the favorite the contestant who has a
probability of winning greater than a half at the Nash equilibrium and
the underdog the contestant whose probability of winning at the Nash
equilibrium is less than a half. Baik (1994, 2004) calls the former the
Nash winner and the latter the Nash loser.
(17) The result is more interesting when we consider the case where
Delegate 2's reservation wage is greater than Delegate 1's.
(18) Santore and Viard (2001) show that the American Bar
Association Model Rules of Professional Conduct that require a minimum
fixed fee of zero--similar to the nonnegativity constraint on
[[beta].sub.i] in this paper--can create economic rents for attorneys
(see Footnote 4). Rasmusen (2001, p. 181) mentions in a standard
principal-agent framework that the principal may pick a contract in
which she pays the agent more than his reservation wage. Schoonbeek
(2002) shows in a two-player contest with unilateral delegation that the
equilibrium expected utility of the delegate may be greater than his
reservation wage. Lawarree and Shin (2005) show that within a flexible
organization, not only an efficient agent but also an inefficient agent
may acquire a rent.
(19) For Type III, ([[alpha].sup.*.sub.1], [[alpha].sup.*.sub.2])
occurs at the intersection of the graphs of [m.sup.0.sub.1] and
[m.sup.0.sub.2], which lies on line segment DS in Figure 3. We obtain
([[alpha].sup.*.sub.1], [[alpha].sup.*.sub.2]) = (4R/[v.sub.1],
4R/[v.sub.2]).
(20) This interpretation makes sense if the prize is a pecuniary one, but it may not make sense if the prize is, for example, winning a
criminal trial.
(21) A lower cap on [[alpha].sub.i] can be justified by the fact
that many states in the United States have limits on contingent fees for
tort cases. See, for example, Danzon (1983), Rubinfeld and Scotchmer
(1993), and Emons (2000).
Table 1. Three Types of Equilibrium--Contracts Pairs
Type I Type II
Reservation 0 [less than or [R.sup.A] [less
wage equal to] R than or equal to]
< [R.sup.A] R < [R.sup.D]
Contracts [[alpha].sup.*.sub.1] [[alpha].sup.*.sub.1] <
< [[alpha].sup. [[alpha].sup.*.sub.2]
*.sub.2] < 1/2 [[beta].sup.*.sub.1] =
[[beta].sup.*.sub.1] = [[beta].sup.*.sub.1] = 0
[[beta].sup.*.sub.1] = 0
Compensation [[alpha].sup.*.sub.1] [[alpha].sup.*.sub.1]
spreads [v.sub.1] > [[alpha] [v.sub.1] >
.sup.*.sub.2] [[alpha].sup.*.sub.2]
[v.sub.2] [v.sub.2]
Delegates' [x.sup.*.sub.1] > [x.sup.*.sub.1] >
effort levels [x.sup.*.sub.2] [x.sup.*.sub.2]
Probability [p.sub.1] [p.sub.1]
of winning ([x.sup.*.sub.1], ([x.sup.*.sub.1],
[x.sup.*.sub.2]) [x.sup.*.sub.2]) > 1/2
> 1/2
Expected
payoffs for
the [[pi].sup.*.sub.1] > [[pi].sup.*.sub.1] >
delegates [[pi].sup.*.sub.2] > R [[pi].sup.*.sub.2] = R
Expected
payoffs for [G.sup.*.sub.1] > [G.sup.*.sub.1] >
the players [G.sup.*.sub.2] > 0 [G.sup.*.sub.2] > 0
Type III
Reservation [R.sup.D] [less
wage than or equal to]
R < [V.sup.2]/4
Contracts [[alpha].sup.*.sub.1] <
[[alpha].sup.*.sub.2]
[[beta].sup.*.sub.1] =
[[beta].sup.*.sub.1] = 0
Compensation [[alpha].sup.*.sub.1]
spreads [v.sub.1] =
[[alpha].sup.*.sub.2]
[v.sub.2]
Delegates' [x.sup.*.sub.1] =
effort levels [x.sup.*.sub.2]
Probability [p.sub.1]
of winning ([x.sup.*.sub.1],
[x.sup.*.sub.2])
= 1/2
Expected
payoffs for
the [[pi].sup.*.sub.1] =
delegates [[pi].sup.*.sub.2] = R
Expected
payoffs for [G.sup.*.sub.1] >
the players [G.sup.*.sub.2] > 0