Do rent-seeking groups announce their sharing rules?
Baik, Kyung Hwan ; Lee, Dongryul
I. INTRODUCTION
Rent seeking can be seen all around us. Firms or coalitions of
firms expend resources to influence government officials who have
authority to award monopoly rights. Self-interested pressure groups
lobby for government subsidies. Foreign producers exert effort to avoid
imposing tariffs or quotas on their products. Interest groups or
lobbyists contribute to the election of politicians who promise to
advance favorable legislation. Research and development (R&D) joint
ventures of firms invest resources to invent new products or new
production technologies that will create economic rents for specified periods of time under government protection. Individual politicians or
political parties campaign hard to win elections.
The literature on rent seeking is enormous and growing. Important
papers in this literature include Tullock (1967, 1980), Krueger (1974),
Posner (1975), Hillman and Katz (1984), Appelbaum and Katz (1987), Dixit
(1987), Hillman and Riley (1989), Hirshleifer (1989), Katz, Nitzan, and
Rosenberg (1990), Ellingsen (1991), Nitzan (1991b), Baik and Shogren
(1992), Leininger (1993), Che and Gale (1997), Konrad and Schlesinger (1997), Hurley and Shogren (1998), Morgan (2003), and Baye and Hoppe
(2003). (1)
Recently, many economists have studied collective rent
seeking--that is, competition for a rent among groups of players in
which the players in each group first decide jointly how to share the
rent among themselves if one of them (or the group) wins it, and then
all the players in the groups simultaneously and independently choose
their effort levels. In this literature, public or private information
is exogenously assumed regarding sharing rules. Baik and Lee (2007) and
Nitzan and Ueda (2007) consider collective rent seeking between groups
in which sharing rules are private information--that is, the players in
each group expend their effort without observing the sharing rule to
which the players in the other groups agreed. All the rest of the
literature considers the case where sharing rules are public
information--that is, the players in each group expend their effort
after observing the sharing rules of the other groups: See, for example,
Nitzan (1991a,b), Balk (1994), Lee (1995), Hausken (1995), Davis and
Reilly (1999), Baik and Lee (2001), Ueda (2002), and Baik et al. (2006).
However, one may well expect that each group has the option of
releasing or not its sharing-rule information. Accordingly, the purpose
of this paper is to study collective rent seeking between two groups in
which the groups first decide independently whether or not to release
their sharing-rule information; then, the groups announce their
decisions simultaneously before choosing their sharing rules. We examine
the groups' decisions on releasing sharing-rule information, their
sharing rules, and the effort levels and payoffs of the individual
players in equilibrium.
We formally consider the following three-stage game. In the first
stage, each group decides and announces whether it will release to the
rival group the information about its sharing rule, which will be
determined in the second stage. In the second stage, the players in each
group jointly choose their sharing rule, and then each group releases
the information about its sharing rule if it decided to do so in the
first stage. In the third stage, all the players in both groups choose
their effort levels simultaneously and independently. At the end of this
third stage, the winning player is chosen, and the winner shares the
prize with the other players in his group according to the sharing rule
on which they agreed in the second stage.
Solving the game, we find the following. First, the case where both
groups release their sharing-rule information never occurs in
equilibrium. Second, the case where neither group releases its
sharing-rule information occurs only if the players are evenly matched.
Third, when the players are unevenly matched, one group releases its
sharing-rule information and the other does not. In this case, if the
players coordinate to attain the Pareto-superior expected payoffs, the
underdog releases its sharing-rule information and the favorite does
not. (2)
This paper is related to Muller and Warneryd (2001), Stein and
Rapoport (2004), Garfinkel (2004), Hausken (2005), and Inderst, Muller,
and Warneryd (2007). Muller and Warneryd (2001) consider a situation in
which managers in a firm jointly produce a surplus, and confront a
costly distributional conflict among themselves over the produced
surplus. They compare two ownership structures, inside ownership and
outside ownership, in several respects. They show that, with outside
ownership, less resources may be wasted and the managers have less
incentive to make firm-specific investments. Stein and Rapoport (2004)
consider two different contest structures, the between-group model and
the semi-finals model. In the between-group model, the groups first
compete with one another to win the prize, and then the players in the
winning group compete against each other to win the prize. In the
semi-finals model, each group first selects the finalist from among its
players, and then the finalists--one for each group--compete to win the
prize. They show that the semi-finals contest structure tends to
generate greater expenditures than the between-group contest structure.
Garfinkel (2004) studies endogenous alliance formation and its effect on
the severity of conflict in a three-stage model of distributional
conflict in which individuals can form alliances in the first stage.
Hausken (2005) considers two distinct but related models: the production
and conflict model in which each agent allocates his resource between
production and fighting, and the rent-seeking model in which each agent
uses his resource only for fighting. He compares the production and
conflict model and the rent-seeking model in several respects. Inderst,
Mailer, and Warneryd (2007) look at influence costs incurred due to
distributional conflict in organizations. They find that influence costs
may be lower in multidivisional organizations than single-tier
organizations.
The paper proceeds as follows. In Section II, we present the model
and set up the game. In Section III, we analyze the four subgames that
start at the second stage of the full game. We obtain the groups'
sharing rules, the players' effort levels, and their expected
payoffs in each subgame. Section IV analyzes the first stage of the full
game--that is, we examine the groups' decisions on releasing
sharing-rule information. In Section V, we select the Pareto-superior
equilibrium in the case where the players are unevenly matched, and
discuss the outcomes in the selected equilibrium. In Section VI, we
discuss a possible commitment problem. Finally, Section VII offers our
conclusions.
II. THE MODEL
Consider a rent-seeking contest (or, in general, a contest) in
which players compete by expending irreversible effort to win a rent or
a prize. Each player belongs to one of two groups, 1 and 2. Each group
consists of n players where n [greater than or equal to] 2, and all the
players are risk-neutral. The players' valuations for the prize may
differ: Each player in group 1 values the prize at [[upsilon].sub.1] and
each player in group 2 values it at [[upsilon].sub.2]. The prize will be
awarded to one of the players. (3) Let [x.sub.ik] represent the effort
level expended by player k in group i, and let [X.sub.i] represent the
effort level expended by all the players in group i, so that [X.sub.i] =
[[summation].sup.n.sub.k=1] [x.sub.ik]. If the total effort level is
positive, the probability that player k in group 1 wins the prize is
given by [p.sub.1k] = [gamma][x.sub.ik]/([gamma][X.sub.1] + [X.sub.2]),
where [gamma] > 0, and the probability that player k in group 2 wins
is given by [p.sub.2k] = [x.sub.2k] / ([gamma][X.sub.1] + [X.sub.2]).
(4) Note that given the total effort level, the probability that a
player wins the prize depends only on his own effort level, not on his
group's effort level. The probability of winning for each player is
equal to 1/2n if all the players expend zero effort. The parameter [gamma] represents abilities of group l's players in the contest
relative to those of group 2's players. For example, if [gamma]
> 1, it means that each player in group 1 has more ability than each
player in group 2--in other words, when they exert the same effort, each
player in group l has a greater probability of winning than each player
in group 2.
The winner "shares" the prize with the other players in
his group. If a player in group i wins the prize, the winning player
takes [[sigma].sub.i][[upsilon].sub.i] and each losing player in the
group "takes" (1 - [[sigma].sub.i])[[upsilon].sub.i]/(n - 1),
where [[upsilon].sub.i] [greater than or equal to] 1/n. (5) We call
[[sigma].sub.i] the winner's fractional share of group i, which the
players in the group agree on before they choose their effort levels.
(6) If [[sigma].sub.i] = 1/n holds, the players in group i share the
prize equally when a player in that group wins it. In the case where l/n
[less than or equal to] [[sigma].sub.i] < 1, the winner takes less
than the prize. When the winner's fractional share is equal to
unity, the winner takes all the prize. In the case where [[sigma].sub.i]
> 1, the winner takes all the prize and further receives
"bounties" from the other players in his group. Thus, in this
case, the winner earns more than the prize.
Let [[pi].sub.ik] represent the expected payoff for player k in
group i. Then the payoff function for player k in group i is
(1)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII],
where [p.sub.ik] is the probability that player k in group i wins
the prize and [[summation].sup.n.sub.j[not equal to]k] [p.sub.ij] is the
probability that any one of the other players in group i wins the prize.
We formally consider the following three-stage game. In the first
stage, each group decides independently whether it will release to the
rival group the information about its sharing rule (or, equivalently,
its winner's fractional share), which will be determined in the
second stage. The groups announce their decisions simultaneously. In the
second stage, the players in each group jointly choose their sharing
rule, and then each group releases the information about its sharing
rule if it decided to do so in the first stage. (7) In the third stage,
all the players in both groups choose their effort levels simultaneously
and independently. At the end of this third stage, the winning player is
chosen, and the winner "shares" the prize with the other
players in his group according to the sharing rule on which they agreed
in the second stage. We assume that there is no transaction cost
associated with negotiating an agreement and enforcing compliance. We
also assume that all the above is common knowledge among the players.
III. THE FOUR SUBGAMES STARTING AT THE SECOND STAGE
We first analyze the subgames that start at the second stage of the
full game. There are four such subgames: the (NR, NR) subgame, the (R,
NR) subgame, the (NR, R) subgame, and the (R, R) subgame, where NR
denotes the action of announcing, in the first stage, that the
sharing-rule information will not be released and R the action of
announcing that it will be released. The (NR, NR) subgame (also called
the no-release subgame) arises when both groups announce that they will
not release their sharing-rule information. If group l announces that it
will release its sharing-rule information but group 2 announces the
opposite, then the (R, NR) subgame arises. The (NR, R) subgame arises
when group 1 announces that it will not release its sharing-rule
information but group 2 announces the opposite. Finally, the (R, R)
subgame (also called the bilateral-release subgame) arises when both
groups announce that they will release their sharing-rule information.
A. The (NR, NR) Subgame
In the no-release subgame, the players in each group first choose
their sharing rule jointly, and then choose their effort levels
simultaneously and independently, without observing the other
group's sharing rule or effort levels. (8) To solve the subgame, we
need to find the groups' sharing rules and the players' effort
levels that satisfy the following two requirements. First, each
player's effort level is optimal given the sharing rule of his own
group and given the effort levels of all the other players. That is,
each player's effort level is a best response to his group's
sharing rule and the effort levels of all the other players. Second,
each group's sharing rule is optimal given the effort levels of the
players in the other group and given the subsequent effort levels of the
players in the group.
To obtain such equilibrium actions--the groups' sharing rules
and the players' effort levels in equilibrium--we begin by deriving the reaction functions for the players in group 1. Working backward, we
first consider the players' decisions on their effort levels. After
observing his group's sharing rule or equivalently [[sigma].sub.1],
player k in group l seeks to maximize his payoff (1) over his effort
level [x.sub.1k], taking the effort levels of all the other players as
given. (9) We focus on the symmetric equilibrium actions. Thus, let
[x.sub.1k] = [x.sub.1] and [x.sub.2k] = [x.sub.2] for all k. Then the
first-order condition for maximizing Equation (1) reduces to
[[gamma].sup.2][n.sup.2][x.sub.2.sub.1] +
(2[gamma][n.sup.2][x.sub.2] - [[gamma].sup.2] [[[sigma].sub.1]n -
1]][[upsilon].sub.1])[x.sub.1]
+ n[x.sub.2](n[x.sub.2] - [gamma][[sigma].sub.1][[upsilon].sub.1] =
0.
Using this equation, we obtain the following reaction function:
(2)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Next, consider the players' decision on their sharing rule.
Because the players expend the same effort level, they have the same
expected payoff: [[pi].sub.1k] = [[pi].sub.1] for all k. The players
seek to maximize
(3)
([[pi].sub.1][x.sub.2]) = [gamma][[upsilon].sub.1][x.sub.1]
([[sigma].sub.1], [x.sub.2]) / n ([gamma][x.sub.1] ([[sigma].sub.1],
[x.sub.2]) + [x.sub.2]) - [x.sub.1] ([[sigma].sub.1], [x.sub.2)
with respect to [[sigma].sub.1], taking group 2's total effort
level [X.sub.2], or rather [x.sub.2], as given. Note that we obtain
Equation (3) by substituting Equation (2) into Equation (l). From the
first-order condition for maximizing Equation (3), we obtain another
reaction function of group 1:
(4) [[sigma].sub.1]([x.sub.2]) = (1 + [n - 1] [square root of
n[x.sub.2] / [gamma][[upsilon].sub.1]]) / n.
Now consider group 2. After observing his group's sharing rule
or equivalently [[sigma].sub.2], player k in group 2 seeks to maximize
his payoff (1) over his effort level [x.sub.2k], taking the effort
levels of all the other players as given. We focus on the symmetric
equilibrium actions. Thus, the first-order condition for maximizing
Equation (1) reduces to
[n.sup.2][x.sup.2.sub.2] + (2[gamma][n.sup.2][x.sub.1] -
[[[sigma].sub.2]n - 1][[upsilon].sub.2])[x.sub.2] + [gamma]n[x.sub.1]
([gamma]n[x.sub.1] - [[gamma].sub.2][v.sub.2]) = 0.
Using this equation, we obtain the following reaction function:
(5)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Next, consider the players' decision on their sharing rule.
Because the players expend the same effort level, they have the same
expected payoff: [[pi].sub.2k] = [[pi].sub.2] for all k. The players
seek to maximize
(6) [[pi].sub.2]([[sigma].sub.2], [x.sub.1]) =
[[upsilon].sub.2][x.sub.2]([sigma].sub.2], [x.sub.1]) / n
([gamma][x.sub.1] + [x.sub.2] ([[sigma].sub.2], [x.sub.1])) -
[x.sub.2]([[sigma].sub.2], [x.sub.1])
with respect to [[sigma].sub.2], taking group l's total effort
level [X.sub.1], or rather [x.sub.1], as given. Note that we obtain
Equation (6) by substituting Equation (5) into Equation (1). From the
first-order condition for maximizing Equation (6), we obtain another
reaction function of group 2:
(7) [[sigma].sub.2]([x.sub.1]) = (1 + [n - 1] [square root of
[gamma]n[x.sub.1] / [[upsilon].sub.2]) / n.
We are now ready to obtain the symmetric equilibrium actions,
denoted by the 2(n + 1)-tuple vector of actions ([[sigma].sup.NR.sub.1]
[x.sup.NR.sub.1], ..., [x.sup.NR.sub.1], [[sigma].sub.NR.sub.2],
[x.sup.NR.sub.2] ..... [x.sup.NR.sub.2]),] by solving the system of four
simultaneous equations, (2), (4), (5), and (7). Substituting Equation
(4) into Equation (2), and Equation (7) into Equation (5), we have
[x.sub.1] ([x.sub.2]) = ([square root of
[gamma]n[[upsilon].sub.1][x.sub.2] - n[x.sub.2]) / n[gamma]
and
[x.sub.2]([x.sub.1]) = ([square root of
[gamma]n[[upsilon].sub.2][x.sub.1]) - [gamma]n[x.sub.1] / n.
By solving this pair of simultaneous equations, we obtain the
players' equilibrium effort levels, [x.sup.NR.sub.1] and
[x.sup.NR.sub.2]. Next, substituting [x.sup.NR.sub.2] into Equation (4),
and [x.sup.NR.sub.1] into Equation (7), we obtain the groups'
equilibrium sharing rules, [[sigma].sup.NR.sub.1] and
[[sigma].sup.NR.sub.2], respectively. Finally, substituting these
equilibrium actions into Equations (3) and (6), we obtain the
players' equilibrium expected payoffs, [[pi].sup.NR.sub.1] and
[[pi].sup.NR.sub.2].
Lemma 1 summarizes the outcomes of the (NR, NR) subgame.
LEMMA 1. (a) In the symmetric equilibrium of the no-release
subgame, group 1 chooses [[sigma].sup.NR.sub.1] =
([gamma][[upsilon].sub.1] + n[[upsilon].sub.2]) / n
([gamma][[upsilon].sub.1] + [[upsilon].sub.2]), and each player in group
1 expends [x.sup.NR.sub.1] =
[gamma][[upsilon].sup.2.sub.1][[upsilon].sub.2] / n
[([gamma][[upsilon].sub.1] + [[upsilon].sub.2]).sup.2]. Group 2 chooses
[[sigma].sup.NR.sub.2] = ([gamma]n[[upsilon].sub.1] + [[upsilon].sub.2])
/ n ([gamma][[upsilon].sub.1] + [[upsilon].sub.2]), and each player in
group 2 expends [x.sup.NR.sub.2] = [gamma][[upsilon].sub.1] +
[[upsilon].sup.2.sub.2]) / n [([gamma][[upsilon].sub.1] +
[[upsilon].sub.2]).sup.2]. (b) The expected payoff for each player in
group 1 and that for each player in group 2 are [[pi].sup.NR.sub.1] =
[[gamma].sup.2][[upsilon].sup.3.sub.1]) / n [([gamma][[upsilon].sub.1] +
[[upsilon].sub.2]).sup.2] and [[pi].sup.NR.sub.2] =
[[upsilon].sup.3.sub.2] / n [([gamma][[upsilon].sub.1] +
[[upsilon].sub.2]).sup.2].
B. The Unilateral-Release Subgames
Consider first the (R, NR) subgame. In this subgame, group 1
releases its sharing-rule information, but group 2 does not. Thus, when
choosing their effort levels, the players in both groups know group
1's sharing rule, but only the players in group 2 know group
2's sharing rule. (10)
We solve this subgame by viewing it as the following two-stage
game. In the first stage, group 1 chooses its sharing rule and announces
it publicly. In the second stage, after observing group 1's sharing
rule, groups 1 and 2 play a simultaneous-move game. That is, the players
in group 1 choose their effort levels, without observing group 2's
sharing rule or effort levels; the players in group 2 choose
sequentially their sharing rule and effort levels without observing
group 1' s effort levels. (11)
To solve this two-stage game, we work backwards. In the second
stage, the players in both groups know group 1's sharing rule,
[sigma]1. We begin by deriving the reaction functions for the players in
group 1. After observing [sigma]1, player k in group 1 seeks to maximize
his payoff (1) over his effort level [x.sub.1k], taking the effort
levels of all the other players as given. We focus on the symmetric
equilibrium actions. Thus, let [x.sub.1k] = [x.sub.1] and [x.sub.2k] =
[x.sub.2] for all k. Then from the first-order condition for maximizing
Equation (1), we obtain the following reaction function:
(8) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Next, consider group 2. The players in group 2 choose sequentially
their sharing rule and effort levels without observing group 1's
effort levels. Taking exactly the same steps as in Subsection A of
Section III, we obtain the reaction functions of group 2:
(9) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
and
(10) [[sigma].sub.2] ([x.sub.1]) = (1 + [n - 1] [square root of
[gamma]n[x.sub.1] / [[upsilon].sub.2] / n.
These reaction functions are the same as those in Equations (5) and
(7). The reason is that knowing [[sigma].sub.1] does not make any
difference because the payoffs to the players in group 2 do not depend
directly on [[sigma].sub.1].
Then, by solving the system of three simultaneous equations, (8),
(9), and (10), we obtain
(11) [x.sub.1]([[sigma].sub.1]) = [gamma]n[[sigma].sup.2.sub.1]
[[upsilon].sup.2.sub.1] [[upsilon].sub.2] / + [([gamma][[upsilon].sub.1]
+ n[[upsilon].sub.2]).sup.2],
[x.sub.2]([[sigma].sub.1]) = [gamma][[sigma].sub.1]
[[upsilon].sub.1] [[upsilon].sub.2] ([gamma][1 - [[sigma].sub.1]n]
[[upsilon].sub.1] + n[[upsilon].sub.2]) / [([gamma] [[upsilon].sub.1] +
n [[upsilon].sub.2]).sup.2],
and
[[sigma].sub.2]([[sigma].sub.1]) = {[gamma][[upsilon].sub.1] (1 +
n[n - 1][[sigma].sub.1]) + n [[upsilon].sub.2]} /
n([gamma][[upsilon].sub.1] + n [[upsilon].sub.2]).
These are the equilibrium effort levels of the players in group 1,
those of the players in group 2, and group 2's equilibrium sharing
rule, respectively, in the second stage.
Next, consider the first stage in which group 1 chooses its sharing
rule. Because the players in group 1 expend the same effort level in the
second stage, we have: [[pi].sub.1k] = [[pi].sub.1] for all k. Having
perfect foresight about [[pi].sub.1]([[sigma].sub.1]), the players in
group 1 choose their sharing rule which maximizes
(12)
[[pi].sub.1]([[sigma].sub.1]) =
[gamma][[upsilon].sub.1][x.sub.1]([[sigma].sub.1]) / n
n([gamma][x.sub.1]([[sigma].sub.1]) + [x.sub.2] ([[sigma].sub.1])) -
[x.sub.1] ([[sigma].sub.1]).
Note that we obtain Equation (12) by substituting [x.sub.1]
([[sigma].sub.1]) and [x.sub.2]([[sigma].sub.1]) in Equation (11) into
Equation (1). From the first-order condition for maximizing Equation
(12) with respect to [[sigma].sub.1], we obtain group 1's
equilibrium sharing rule, [[sigma].sup.1R.sub.1].
Now, substituting [[sigma].sup.1R.sub.1] into
[x.sub.i]([[sigma].sub.1]), [x.sub.2]([[sigma].sub.1]), and
[[sigma].sub.2]([[sigma].sub.1]) in Equation (11), we obtain the
players' equilibrium effort levels, [x.sup.1R.sub.1] and
[x.sup.1R.sub.2], and group 2's equilibrium sharing rule,
[[sigma].sup.1R.sub.2], respectively. Next, using these equilibrium
actions, we obtain the players' equilibrium expected payoffs,
[[pi].sup.1R.sub.1] and [[pi].sup.1R.sub.2].
Lemma 2 summarizes the outcomes of the (R, NR) subgame.
LEMMA 2. (a) In the symmetric equilibrium of the (R, NR) subgame,
group 1 chooses [[sigma].sup.1R.sub.1] = ([gamma][[upsilon].sub.1] +
n[[upsilon].sub.2]) / 2n[[upsilon].sub.2], and each player in group 1
expends [x.sup.1R.sub.1] = [gamma][[upsilon].sup.2.sub.1] /
4n[[upsilon].sub.2]. Group 2 chooses [[sigma].sup.1R.sub.2] = ([gamma][n
- 1][[upsilon].sub.1] + 2[[upsilon].sub.2])/2n[[upsilon].sub.2], and
each player in group 2 expends [x.sup.1R.sub.2] =
[gamma][[upsilon].sub.1] (2[[upsilon].sub.2] - [gamma][[upsilon].sub.1])
/ 4n[[upsilon].sub.2]. (b) The expected payoff for each player in group
1 and that for each player in group 2 are [[pi].sup.1R.sub.1] =
[gamma][[upsilon].sup.2.sub.1] / 4n[[upsilon].sub.2] and
[[pi].sup.1R.sub.2] = [(2[[upsilon].sub.2] -
[gamma][[upsilon].sub.1]).sup.2] / 4n[[upsilon].sub.2].
Next, consider the other unilateral-release subgame, the (NR, R)
subgame. In this subgame, group 2 releases its sharing-rule information,
but group 1 does not. Thus, when choosing their effort levels, the
players in both groups know group 2's sharing rule, but only the
players in group 1 know group l's sharing rule.
We solve this subgame by viewing it as the following two-stage
game. In the first stage, group 2 chooses its sharing rule and announces
it publicly. In the second stage, after observing group 2's sharing
rule, groups 1 and 2 play a simultaneous-move game. That is, the players
in group 1 choose sequentially their sharing rule and effort levels
without observing group 2's effort levels; the players in group 2
choose their effort levels, without observing group l's sharing
rule or effort levels.
Because the analysis is similar to that for the (R, NR) subgame, we
only report the results. Lemma 3 summarizes the outcomes of the (NR, R)
subgame.
LEMMA 3. (a) In the symmetric equilibrium of the (NR, R) subgame,
group 1 chooses [[sigma].sup.2R.sub.1] = (2[gamma][v.sub.1] + [n -
1][v.sub.2])/2[gamma][v.sub.1], and each player in group 1 expends
[x.sup.2R.sub.1] = [v.sub.2](2[gamma][v.sub.1] -
[v.sub.2])/4[[gamma].sub.2]n[v.sub.1]. Group 2 chooses
[[sigma].sup.2R.sub.2]= ([gamma]n[v.sub.1] +
[v.sub.2])/2[gamma]n[v.sub.1], and each player in group 2 expends
[x.sup.2R.sub.2] = [v.sup.2.sub.2] /4[gamma]n[v.sub.1]. (b) The expected
payoff for each player in group 1 and that for each player in group 2
are [[pi].sup.2R.sub.1] = [(2[gamma][v.sub.1] -
[v.sub.2]).sup.2]/4[[gamma].sup.2]n[v.sub.1] and [[pi].sub.2R.sub.2] =
[v.sup.2.sub.2]/4[gamma][v.sub.1].
C. The (R, R) Subgame
The bilateral-release subgame has two stages. In the first stage,
the players in each group jointly choose their sharing rule and announce
it publicly. In the second stage, after observing the sharing rules, the
players in both groups choose their effort levels simultaneously and
independently.
To solve for a subgame-perfect equilibrium of this subgame, we work
backwards. In the second stage, the players in both groups know the
groups' sharing rules, [[sigma].sub.1] and [[sigma].sub.2]. We
begin by deriving the reaction functions for the players in group 1.
Player k in group 1 seeks to maximize his payoff (1) over his effort
level [x.sub.1k], taking the effort levels of all the other players as
given. We focus on the symmetric equilibrium. Thus, let [x.sub.1k] =
[x.sub.1] and [x.sub.2k] = [x.sub.2] for all k. Then from the
first-order condition for maximizing Equation (1), we obtain the
following reaction function:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Similarly, the reaction function for each player in group 2 is
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Using these reaction functions, we obtain the symmetric Nash
equilibrium in the second stage of the bilateral-release subgame (12):
(13) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Let [[pi].sub.i] ([[sigma].sub.1], [[sigma].sub.2]) be the expected
payoff of each player in group i at the symmetric Nash equilibrium of
the second stage. Substituting [x.sub.1] ([[sigma].sub.1],
[[sigma].sub.2]) and [x.sub.2] ([[sigma].sub.1], [[sigma].sub.2]) into
Equation (1), we obtain
(14)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Next, consider the first stage in which both groups choose their
sharing rules. In this stage, each player has perfect foresight about
both [[pi].sub.1]([[sigma].sub.1, [[sigma].sub.2]) and
[[pi].sub.2]([sigma].sub.1], [[sigma].sub.2]). Given the other
group's sharing rule, the players in group i choose their sharing
rule that maximizes [[pi].sub.i]([[sigma].sub.1,[[sigma].sub.2). From
the first-order condition for maximizing
[[pi].sub.i]([sigma].sub.1],[[sigma].sub.2]) for i = 1, 2, we obtain the
following reaction functions, [[sigma].sub.1] ([[sigma].sub.2]) for
group 1 and [[sigma].sub.2]([[sigma].sub.1] for group 2, respectively:
(15)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Using these two reaction functions, we obtain the groups'
sharing rules, [[sigma].sup.BR.sub.1] and [[sigma].sup.BR.sub.2], which
are specified in the subgame-perfect equilibrium of the
bilateral-release subgame:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Now, substituting [[sigma].sup.BR.sub.1] and [[sigma].sup.BR.sub.2]
into [x.sub.1] ([[sigma].sub.1], [[sigma].sub.2]) and [x.sub.2]
([[sigma].sub.1], [[sigma].sub.2]) in Equation (13), we obtain the
players' equilibrium effort levels, [x.sup.BR.sub.1] and
[X.sup.BR.sub.2], respectively. Next, using these equilibrium actions,
we obtain the players' equilibrium expected payoffs,
[[pi].sup.BR.sub.1] and [[pi].sup.BR.sub.2].
Let Q [equivalent to]
[[gamma].sup.2]n[v.sup.2.sub.1]+n[v.sup.2.sub.2] -
[gamma][v.sub.1][v.sub.2], G [equivalent to] [gamma]n + 1)[v.sub.1] -
n[v.sub.2], D [equivalent to]
[n.sup.2][([gamma][v.sub.1]+[v.sub.2]).sup.2], and H [equivalent to] (n
+ 1)[v.sub.2] - [gamma]n[v.sub.1]. Lemma 4 summarizes the outcomes of
the (R, R) subgame.
LEMMA 4. (a) In the symmetric equilibrium of the (R, R) subgame,
group 1 chooses [[sigma].sup.BR.sub.1], and each player in group 1
expends [x.sup.BR.sub.1] = Q x G/[gamma]D. Group 2 chooses
[[sigma].sup.BR.sub.2], and each player in group 2 expends
[X.sup.BR.sub.2] = Q x H/D. (b) The expected payoff for each player in
group 1 and that for each player in group 2 are [[pi].sup.BR.sub.1] =
[v.sub.2] x [G.sup.2]/[gamma]D and [[pi].sub.BR.sub.2] =
[gamma][v.sub.1] x [H.sup.2]/D.
IV. GROUPS' DECISIONS ON RELEASING SHARING-RULE INFORMATION
Now consider the first stage of the full game. In this stage, each
group first decides independently whether or not to release its
sharing-rule information to the rival group, and then both groups
announce their decisions simultaneously. In short, each group chooses
one of the following two actions: announcing that it will not release
its sharing-rule information or announcing that it will release the
information. Recall that we denote the former action by NR and the
latter one by R. We then have four possible combinations of the
groups' actions: (NR, NR), (R, NR), (NR, R), and (R, R). For
example, if group 1 chooses NR and group 2 chooses R in the first stage,
then the combination (NR, R) arises. Table 1 shows the expected payoffs
of the players for the four possible combinations. The combination (NR,
NR) leads to the no-release subgame analyzed in Subsection A of Section
III, so that the expected payoff for each player in group 1 is
[[pi].sup.NR.sub.1], and that for each player in group 2 is
[[pi].sup.NR].sub.2] (see Lemma 1). Similarly, we have
[[pi].sup.1R.sub.1] and [[pi].sup.1R.sub.2] for the combination (R, NR),
[[pi].sup.2R.sub.1] and [[pi].sup.2R.sub.2] for the combination (NR, R),
and [[pi].sup.BR.sub.1] and [[pi].sup.BR.sub.2] for the combination (R,
R), which come from Lemmas 2, 3, and 4, respectively.
Which combinations occur in the equilibria of the full game?
Looking at Table 1 that illustrates the strategic interaction between
the groups in the first stage, we have the following. If
[[pi].sup.NR.sub.1] [greater than or equal to] [[pi].sup.1R.sub.1] and
[[pi].sup.NR.sub.2] [greater than or equal to] [[pi].sup.2R.sub.2], then
the combination (NR, NR) occurs in equilibrium; if [[pi].sup.1R.sub.1]
[greater than or equal to] [[pi].sup.NR.sub.1]_ and [[pi].sup.1R.sub.2]
[[greater than or equal to] [[pi].sup.BR.sub.2], then the combination
(R, NR) arises; if [[pi].sup.2R.sub.1] [greater than or equal to]
[[pi].sup.BR.sub.1] and [[pi].sup.2R.sub.2] [greater than or equal to]
[[pi].sup.NR.sub.2], then the combination (NR, R) arises; if
[[pi].sup.BR.sub.1] [greater than or equal to] [[pi].sup.2R.sub.1] and
[[pi].sup.BR.sub.2] [greater than or equal to] [[pi].sup.1R.sub.1] then
the combination (R, R) occurs in equilibrium. Using these statements
together with Lemmas 1 through 4, we obtain Proposition 1. For concise
exposition, from this point on, we let [v.sub.1] [equivalent to]
[alpha][v.sub.2], where [alpha] > 0. (13)
PROPOSITION 1. (a) If [alpha][gamma] = 1, then the combinations,
(R, NR), (NR, R), and (NR, NR), occur in the equilibria of the full
game. (b) If [alpha][gamma] [not equal to] 1, then the combinations, (R,
NR) and (NR, R), occur in the equilibria of the full game.
When [alpha][gamma] = 1, all the players in the contest have the
same "composite strength"--strength determined by their
valuations for the prize as well as their relative abilities. In this
case, if they were to compete individually to win the prize by exerting
effort simultaneously, they would have the same probability of winning
in equilibrium. When [alpha][gamma] = 1, we have ([[pi].sup.1R.sub.1],
[[pi].sup.1R.sub.2]) = [[pi].sup.2R.sub.1],[[pi].sup.2R.sub.2] =
([[pi].sup.NR.sub.1], [[pi].sup.NR.sub.2]) = ([alpha][v.sub.2]/4n,
[v.sub.2]/4n) and ([[pi].sup.BR.sub.1], [[pi].sup.BR.sub.2) =
([alpha][v.sub.2]/4[n.sup.2], [v.sub.2]/4[n.sup.2]). This implies that,
given group 2's action NR, each player in group. 1 is indifferent between NR and R because [[pi].sup.NR.sub.1 = [[pi].sup.1R.sub.1]; given
group 1's action NR, each player in group 2 is also indifferent
between NR and R because [[pi].sup.NR.sub.2] = [[pi].sup.2R.sub.2];
furthermore, the three combinations--(R, NR), (NR, R), and (NR,
NR)--lead to the same pair of expected payoffs.
Proposition 1 immediately implies that the combination (R,
R)--which leads to the Paretoinferior pair of expected payoffs,
([[pi].sup.BR.sub.1], [[pi].sup.BR.sub.2])--never occurs in equilibrium.
(14) This result is very surprising because all the papers except Baik
and Lee (2007) that study collective rent seeking assume that sharing
rules are public information. Moreover, it is surprising because this
type of game tends to yield (weakly) dominant actions for the
"players" in the first stage which lead to a Pareto-inferior
equilibrium. (15) The result can be explained as follows. Given the
rival group's action R, group i has two choices, R or NR. If the
group chooses R, then both groups will announce their sharing rules. In
this case, each group will choose a large winner's fractional share
to gain strategic advantage against its rival group in the
effort-expending stage. The large winner's fractional shares in
turn will motivate the players to expend large effort levels, which will
result in significantly small expected payoffs to the players in group
i.
On the other hand, if group i chooses NR, then only the rival group
will announce its sharing rule exercising strategic leadership. In this
case, sizing up the rival group's sharing rule, group i will choose
a sharing rule with which it can avoid a big fight against the rival
group. Consequently, the players will expend moderate effort levels,
which will result in sizable expected payoffs to the players in group i.
Hence, the players in group i choose NR instead of choosing R.
Proposition 1 says that the equilibrium involving (R, NR) and the
equilibrium involving (NR, R) always occur regardless of the value of
[alpha][gamma]. That is, given the rival group's action R, group i
has no incentive to deviate from its action NR; given the rival
group's action NR, group i has no incentive to deviate from its
action R. The former statement is supported by the explanations in the
preceding paragraph. The latter is supported by the following intuitive
explanation. With the rival group's action NR and its own action R,
group i enjoys a first-mover advantage by announcing its sharing rule
before the rival group chooses its sharing rule. However, if group i
chooses NR instead of R, it loses the strategic leadership and plays the
simultaneous-move game with sequential moves against the rival group,
which results in smaller or equal expected payoffs to the players in
group i. Thus, group i has no incentive to deviate from its action R.
Another interesting result is that, in equilibrium, there exist the
strategical leader and the strategical follower that are determined
endogenously. Indeed, in the equilibria--except the one involving the
combination (NR, NR)--one group chooses R and the other chooses NR; the
former group becomes the leader and the latter one becomes the follower.
V. THE UNDERDOG BECOMES THE STRATEGICAL LEADER
Now a natural and interesting question is: Which group chooses R
and becomes the leader in equilibrium? At a first glance of Proposition
1, this question seems to make no sense because the equilibrium
involving (R, NR) and the equilibrium involving (NR, R) always exist
together. However, the question does make sense because we can narrow
the equilibrium set. Below we narrow the equilibrium set in the case
where [alpha][gamma] [not equal to] 1. Using Lemmas 2 and 3, we obtain
Lemma 5.
LEMMA 5. (a) If [alpha][gamma] > 1, then the expected payoffs of
the players are greater in the equilibrium involving (NR, R) than in the
equilibrium involving (R, NR): [[pi].sup.2R.sub.i] >
[[pi].sup.1R.sub.i] for i = 1,2. (b) If [alpha][gamma] < 1, then the
expected payoffs of the players are greater in the equilibrium involving
(R, NR) than in the equilibrium involving (NR, R): [[pi].sup.1R.sub.i]
> [[pi].sup.2R.sub.i] for i = 1, 2.
Following Dixit (1987), we call group 1 the favorite and group 2
the underdog, when [alpha][gamma] > 1; we call group 1 the underdog
and group 2 the favorite, when [alpha][gamma] < 1.
Lemma 5 says that, if [alpha][gamma] > 1, then the combination
(NR, R) leads to a Pareto-superior pair of expected payoffs, compared
with the combination (R, NR); if [alpha][gamma] < 1, then the
opposite holds true. This implies that the expected payoffs of the
players are greater in the equilibrium in which the underdog chooses R
and the favorite chooses NR--thus, the underdog becomes the strategical
leader. How do we explain this? Consider, for example, the case where
[alpha][gamma] > 1 and thus group 2 is the underdog. In this case, we
have 0.5< [[sigma].sup.2R.sub.1] < [[sigma].sup.1R.sub.1] < 1
and 0.5< [[sigma].sup.2R.sub.2] < [[sigma].sup.1R.sub.2] < 1
(see Lemmas 2 and 3). That is, both groups choose smaller winner's
fractional shares in the equilibrium involving (NR, R)--in which the
underdog is the leader--than in the equilibrium involving (R, NR). The
smaller winner's fractional shares in turn cause the players to
expend smaller effort levels, which result in larger expected payoffs to
the players, compared with the equilibrium involving (R, NR). (16) The
remaining question is then: Why do the groups choose smaller
winner's fractional shares when the underdog is the leader? We
answer this question as follows. When the underdog is the leader, the
underdog restrains itself to avoid stiff competition against the strong
rival. This in turn allows the favorite to ease up and respond
efficiently. On the other hand, when the favorite is the leader, it
preempts by choosing a large winner's fractional share. In response
to this preemptive behavior, the underdog follows suit--facing
aggressive players in the rival group, the players in the underdog group
must make themselves aggressive by choosing a large winner's
fractional share.
Now, using Lemma 5, we can narrow the equilibrium set. Let us
assume that the groups can coordinate each other in choosing their
actions in the first stage. Then we expect that the groups, or rather
the players, will end up with the Pareto-superior expected payoffs. This
means that if [alpha][gamma] > 1, then group 1 chooses NR and group 2
chooses R; if [alpha][gamma] < 1, then group 1 chooses R and group 2
chooses NR in the first stage. Proposition 2 highlights this result.
PROPOSITION 2. If [alpha][gamma] [not equal to] 1, then the
underdog chooses R--that is, the underdog announces that it will release
its sharing-rule information-and the favorite chooses NR in the first
stage. Thus, the underdog becomes the strategical leader and the
favorite becomes the strategical follower. (17)
Table 2 presents the outcomes of the contest in the selected
equilibrium. It says that the favorite chooses a smaller winner's
fractional share than the underdog. (18) This happens because the
favorite eases up, possessing a competitive advantage over the underdog,
whereas the players in the underdog group motivate themselves by
choosing a more "selfish" sharing rule to overcome their
competitive disadvantage. Table 2 also says that the equilibrium
winner's fractional shares are less than unity. This means that, if
a player in a group wins the prize, the winner helps the losers in his
group. Finally, Table 2 says that each player in the favorite group has
a greater probability of winning than each player in the underdog group.
(19) Note that, in Table 2, [p.sup.2R.sub.i] represents the probability
of winning for each player in group i in the equilibrium involving (NR,
R); and [p.sup.1R.sub.i]] represents the probability of winning for each
player in group i in the equilibrium involving (R, NR).
Of great interest is to compare the total effort level and the
expected payoffs in the selected equilibrium with those obtained in the
other three subgames--which start at the second stage of the full
game--that are not specified in the selected equilibrium. Comparing the
total effort level in the selected equilibrium with that obtained in the
(NR, NR) subgame, and with that obtained in the (R, R) subgame, we find
that the players expend less effort in the selected equilibrium than in
each of the two subgames: In terms of symbols, if [alpha][gamma] > 1
then we have [X.sup.2R]+ [X.sub.2R.sub.2] < [X.sup.NR.sub.1] +
[X.sup.NR.sub.2] and [X.sub.2R.sub.1] + [X.sup.2R.sub.2] <
[X.sup.BR.sub.1] + [X.sub.BR.sub.2]; if [alpha][gamma] > 1, then we
have [X.sup.1R.sub.1] + [X.sup.1R.sub.2] < [X.sup.NR.sub.1] +
[X.sup.NR.sub.2] and [X.sup.1R.sub.1] + [X.sup.1R.sub.2] <
[X.sup.BR.sub.1] + [X.sup.BR.sub.2]. (20) This together with footnote 16
demonstrates that total effort level (or the extent of rent dissipation)
is minimized in the selected equilibrium. In other words, the social
costs associated with collective rent seeking is minimized in the
selected equilibrium. (21) Next, comparing the expected payoffs in the
selected equilibrium with those obtained in the (NR, NR) subgame, and
with those obtained in the (R, R) subgame, we find that the expected
payoff for each player is greater in the selected equilibrium than in
each of the two subgames. More specifically, if [alpha][gamma] > 1,
then we have [[pi].sup.BR.sub.i] < [[pi].sup.NR.sub.i] <
[[pi].sup.2R.sub.i] for i = 1, 2; if [alpha][gamma] < 1, then we have
[[pi].sup.BR.sub.i] < [[pi].sup.NR.sub.i] < [[pi].sup.1R.sub.i]
for i = 1, 2. This together with Lemma 5 demonstrates that the expected
payoff for each player is maximized in the selected equilibrium. To sum
up, in the selected equilibrium, the extent of rent dissipation is
minimized, and each player's expected payoff is maximized, compared
with the other three subgames that are not specified in the selected
equilibrium. (22) On the basis of this, we argue that it benefits both
the society and rent seekers to let rent-seeking groups decide freely on
releasing their sharing-rule information. Furthermore, we argue that, to
reduce the extent of rent dissipation, policies or regulations or
institutions that require rent-seeking groups to release or hide their
sharing-rule information should not be enacted or established; but those
that facilitate rent seekers or rent-seeking groups to coordinate each
other in choosing their actions and those that facilitate them to commit
to their chosen actions may be established.
VI. A POSSIBLE COMMITMENT PROBLEM
So tar we have abstracted away from the possibility that a group
reneges on its first-stage decision on releasing sharing-rule
information. However, a group may renege on its first-stage decision if
it can do so. For example, recall from Section V that, in the selected
equilibrium, the underdog chooses R and the favorite chooses NR in the
first stage--that is, the favorite announces in the first stage that it
will not release its sharing-rule information. However, after observing
the underdog's equilibrium sharing rule, the favorite may publicly
announce--if it can do--its "aggressive" sharing rule, which
is the best response to the underdog's "equilibrium"
sharing rule. Indeed, in a specific example below, the favorite (or
group 1) has an incentive to announce its "aggressive" sharing
rule.
Consider the case where [alpha][gamma] > 1. In this case, group
1 chooses NR and group 2 chooses R in the selected equilibrium--that is,
the (NR, R) subgame occurs in the selected equilibrium. Then, using
Lemma 3 and that [v.sub.1] [equivalent to] [alpha][v.sub.2], we obtain
(16) [[sigma].sup.2R.sub.1] = (2[alpha][gamma] + n -
1)/2[alpha][gamma]n,
[[sigma].sup.2R.sub.2] = ([alpha][gamma]n + 1)/2[alpha][gamma]n,
and
[[pi].sup.2R.sub.1] = [[upsilon].sub.2][(2[alpha][gamma] -
1).sup.2]/4[alpha][[gamma].sup.2]n,
where [[sigma].sup.2R.sub.1] is the equilibrium sharing rule (or,
equivalently, the equilibrium winner's fractional share) of group
1, [[sigma].sup.2R.sub.2] is that of group 2, and [[pi].sup.2R.sub.1] is
the expected payoff for each player in group 1 in the selected
equilibrium.
Suppose now that group 1 can and does renege on its first-stage
decision on releasing sharing-rule information. Specifically, suppose
that, after observing group 2's equilibrium sharing rule
[[sigma].sup.2R.sub.2] in Equation (16), the players in group 1 announce
their sharing rule publicly before choosing their effort levels. Then,
group 1 chooses and announces its "aggressive" sharing rule
[[sigma].sup.d2R.sub.1] instead of choosing and announcing its
"equilibrium" sharing rule [[sigma].sup.2R.sub.1] in Equation
(16), where [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. Note
that we obtain [[sigma].sup.d2R.sub.1] by substituting
[[sigma].sup.2R.sub.2] in Equation (16) into group 1's reaction
function, [[sigma].sub.1] ([[sigma].sub.2]) in Equation (15), in the (R,
R) subgame. Note also that [[sigma].sup.d2R.sub.1] is greater than
[[sigma].sup.2R.sub.1] in Equation (16). Next, substituting
[[sigma].sup.d2R.sub.1] and [[sigma].sup.2R.sub.2] into [x.sub.1]
([[sigma].sub.1], [[sigma].sub.2]) and [x.sub.2] ([[sigma].sub.1],
[[sigma].sub.2]) in Equation (13), then substituting [x.sub.1]
([[sigma].sup.d2R.sub.1], [[sigma].sup.2R.sub.2]) and [x.sub.2]
{[[sigma].sup.d2R.sub.1], [[sigma].sup.2R.sub.2]) into Equation (14),
and using that [[upsilon].sub.1] [equivalent to]
[alpha][[upsilon].sub.2], we obtain the expected payoff for each player
in group 1 from the "deviation":
(17)
[[pi].sup.2R.sub.1] [equivalent to] [[sigma].sup.d2R.sub.1],
[[sigma].sup.2R.sub.2]) = [[upsilon].sub.2] [(2[alpha][gamma] -
1).sup.2] [([alpha][gamma] + 1).sup.2] /
16[[alpha].sup.2][[gamma].sup.3][n.sup.2].
Now, comparing the expected payoffs, [[pi].sup.2R.sub.1] in
Equation (16) and [[pi].sup.d2R.sub.1] in Equation (17), we obtain that
[[pi].sup.2R.sub.1] < [[pi].sup.d2R.sub.1]. This implies that group
1, or rather each player in group 1, would be better off by announcing
its "aggressive" sharing rule rather than by keeping its
first-stage "promise." In other words, it implies that group 1
reneges on its first-stage action if it can do so.
In the example above, if group 1 is not committed to its
first-stage action NR, then group 2 believes that group 1 will announce
its sharing rule rather than keep its first-stage "promise."
This leads to the outcomes of the (R, R) subgame, not to those of the
(NR, R) subgame, and thus the players in both groups are worse off. In
short, if group 1 cannot commit to its first-stage action NR, then a
commitment problem arises. (23)
Clearly, both groups prefer group 1 to be committed to its
first-stage action NR. An immediate, natural question is then: How is
group 1 committed to that action? Or, in general, how are rent-seeking
groups committed to such actions or decisions? We can think of several
commitment devices or ways of their being committed. First, the contest
organizer, if any, or the person who has authority to select the winner
can simply require the groups to keep their "promises."
Second, the contest organizer, the decision-maker, or the rent-seeking
groups can create institutions or make rules to solve the commitment
problem. Third, the rent-seeking groups themselves can have incentives
to maintain their reputations for keeping their "promises."
Fourth, culture can be a commitment device. For example, feelings of
guilt can provide psychological incentives for the groups to keep their
"promises." Finally, even though rent-seeking groups are not
committed to their earlier actions or decisions, a group may not wish to
start a "war" that will devastate itself as well as its rival
groups. In the specific example above, if group 1 announces its
"aggressive" sharing rule after observing group 2's
equilibrium sharing rule, then group 2 may update and re-announce its
sharing rule; then, group 1 may update and re-announce its sharing rule;
and so on. This leads to lower payoffs for the players in both groups,
as compared with the case where group 1 keeps its first-stage
"promise."
VII. CONCLUSIONS
We considered collective rent seeking between two groups in which
each group has the option of releasing or not its sharing-rule
information. More specifically, we considered the following three-stage
game. In the first stage, each group decides and announces whether it
will release to the rival group the information about its sharing rule,
which will be determined in the second stage. In the second stage, the
players in each group jointly choose their sharing rule, and then each
group releases the information about its sharing rule if it decided to
do so in the first stage. In the third stage, all the players in both
groups choose their effort levels simultaneously and independently.
In Section IV, we demonstrated that the case where both groups
release their sharing-rule information never occurs in equilibrium; the
case where neither group releases its sharing-rule information occurs
only if the players are evenly matched. This result is very surprising
because almost all the papers in the literature on collective rent
seeking assume that sharing rules are public information. We also
demonstrated that, when the players are unevenly matched, one group
releases its sharing-rule information and the other does not. Because
the group that releases its sharing-rule information assumes the
leadership role, the former group becomes the strategical leader and the
latter becomes the strategical follower--the roles are determined
endogenously.
In Section V, we first showed that the expected payoffs of the
players are greater in the equilibrium in which the underdog--the group
with "weaker" players--releases its sharing-rule information
and the favorite does not, compared with the equilibrium in which the
favorite releases its sharing-rule information and the underdog does
not. Then, assuming that the players coordinate to attain the
Pareto-superior expected payoffs, we selected an equilibrium that is
Pareto superior to the other when the players are unevenly matched. In
this selected equilibrium, the underdog releases its sharing-rule
information, and the favorite does not; thus the underdog becomes the
strategical leader, and the favorite becomes the strategical follower.
In the selected equilibrium, total effort level (or the extent of rent
dissipation) is minimized, and each player's expected payoff is
maximized, compared with the other three subgames that are not specified
in the selected equilibrium. On the basis of this, we argue that it
benefits both the society and rent seekers to let rent-seeking groups
decide freely on releasing their sharing-rule information.
We considered collective rent seeking in which the prize is awarded
to one of the players, and the winner shares the prize with the other
players in his group. Instead, we can consider collective rent seeking
in which the prize is awarded to one of the groups, and the players in
the winning group share the prize among themselves. As mentioned in
footnote 3, with the sharing rule specification therein, we obtain
exactly the same results.
We assumed that both groups consist of the same number of players.
What happens if we assume that the groups have different numbers of
players? Let us assume that group 1 consists of [n.sub.1] players and
group 2 consists of [n.sub.2] players. First, we can explicitly obtain
the results corresponding to Lemmas 1 through 4, some of which involve
very long mathematical expressions. Second, part (a) of Proposition 1
holds true, regardless of the values of [n.sub.1] and [n.sub.2]. Third,
because of computational complexity involved, we cannot show explicitly
that part (b) of Proposition 1 holds true. (24) However, on the basis of
our experience of finding the equilibria of the game using different
numerical values of the parameters, we believe that it holds true
subject to the constraint corresponding to that specified in footnote
13. (25) Finally, once part (b) of Proposition 1 holds true, then so do
Lemma 5 and Proposition 2. However, it is not generally possible to
compare the total effort level in the selected equilibrium with those
obtained in the other three subgames that are not specified in the
selected equilibrium.
It would be interesting to study the following three-stage game and
compare its outcomes with those obtained in this paper. In the first
stage, the players in each group jointly choose and commit to their
sharing rule. In the second stage, each group decides independently
whether it will release to the rival group the information about its
sharing rule. Each group then releases the information about its sharing
rule if it decided to do so. In the third stage, all the players in both
groups choose their effort levels simultaneously and independently. It
would also be interesting to examine the groups' decisions on
releasing sharing-rule information in a production and conflict model in
which each player allocates his resource between production and
fighting. We leave these for future research.
ABBREVIATION
R&D: Research and Development
doi:10.1111/j.1465-7295.2009.00280.x
APPENDIX
The Game between Two Parties in Which Both Parties' First
Moves Are Not Public Information
Consider a game between two parties, 1 and 2, in which each party
has two sequential moves. The first move of one party is observed by all
the players in both parties before the second moves of the parties are
chosen. However, the first move of the other party is observed only by
the players in that party; it is hidden from the players in the rival
party. The second moves of the parties are chosen simultaneously. For
expositional convenience, the player or group of players in party i, for
i = 1, 2, who chooses the first move is called leader i, and that
choosing the second move is called follower i. We allow leader i and
follower i to be the same player or group of players.
We formally consider the following game. First, leaders 1 and 2
choose actions [a.sub.1] and [a.sub.2] from [A.sub.1] and [A.sub.2],
respectively, where [A.sub.i] denotes the set of all actions available
to leader i. Next, follower 1 observes the action al chosen by leader 1,
but cannot observe the action [a.sub.2] chosen by leader 2. Follower 2,
however, observes both [a.sub.1] and [a.sub.2]. Finally, followers 1 and
2 simultaneously choose actions [b.sub.1] and [b.sub.2] from [B.sub.1]
and [B.sub.2], respectively, where [B.sub.i] is follower i's set of
actions. Let [u.sub.i] represent the (expected) payoff for leader i and
[[upsilon].sub.i] that for follower i. The payoff function for leader i
and that for follower i are given by [u.sub.i] = [u.sub.i]([a.sub.i],
[b.sub.i], [b.sub.j]) and [[upsilon].sub.i] = [[upsilon].sub.i]
([a.sub.i], [b.sub.i], [b.sub.j]), respectively. (Throughout the paper,
when we use i and j at the same time, we mean that i [not equal to] j.)
Note that [a.sub.j] is absent in these functions. This implies that the
payoffs to the players in each party do not depend directly on the first
move of the other party. We assume that all of the above is common
knowledge among the leaders and followers.
The right way to look at this game is that the two parties play the
following two-stage game. In the first stage, leader 1 chooses her
action. In the second stage, after observing leader 1's action,
follower 1 and party 2 play a simultaneous-move game: Follower 1 chooses
his action without observing party 2's sequential actions, and
party 2--specifically, leader 2 and follower 2--chooses its two
sequential actions without observing follower 1's action. In this
way of solving the game, leader 1 is treated as the strategical leader
who moves even before leader 2 moves, whatever the chronological timing
of their decisions may be in the original game.
Equilibrium Actions
To solve the two-stage game, we take the following three steps.
First, we analyze the subgames that start at the second stage of the
game. In each of the subgames, follower 1 and party 2 play a
simultaneous-move game, knowing the action of leader 1 which gave rise
to the subgame. To solve the subgames, we need to find a triple vector,
([b.sup.*.sub.1]([a.sub.1]), [a.sup.*.sub.2]([a.sub.1]),
[b.sup.*.sub.2]([a.sub.1])), which satisfies the following two
requirements. First, for any action al of leader 1, follower 1's
action [b.sup.*.sub.1]([a.sub.1]) is optimal given the action
[b.sup.*.sub.2]([a.sub.1]) of follower 2; follower 2's action
[b.sup.*.sub.2]([a.sub.1]) is optimal given the action
[a.sup.*.sub.2]([a.sub.1]) of leader 2 and given the action
[b.sup.*.sub.1]([a.sub.1]) of follower 1. That is, for any action
[a.sub.1] of leader 1, follower 1's action
[b.sup.*.sub.1]([a.sub.1]) is a best response to follower 2's
action [b.sup.*.sub.2]([a.sub.1]); follower 2's action
[b.sup.*.sub.2]([a.sub.1]) is a best response to leader 2's action
[a.sup.*.sub.2]([a.sub.1]) and follower 1's action
[b.sup.*.sub.1]([a.sub.1]). Second, for any action [a.sub.1] of leader
1, leader 2's action [a.sup.*.sub.2]([a.sub.1]) is optimal given
the action [b.sup.*.sub.1]([a.sub.1]) of follower 1 and given the
subsequent behavior of follower 2. Note that such a triple vector
permits the interpretation that for any action [a.sub.1] of leader 1,
[b.sup.*.sub.1]([a.sub.1]) is a best response to
([a.sup.*.sub.2]([a.sub.1]), [b.sup.*.sub.1]([a.sub.1])) and vice versa.
Next, we analyze the first stage in which leader 1 chooses her
action. We need to find leader 1's action [a.sup.*.sub.1] which is
optimal given the strategy [a.sup.*.sub.2]([a.sub.1]) of leader 2, the
strategy [b.sup.*.sub.1]([a.sub.1]) of follower 1, and
[b.sup.*.sub.2]([a.sub.1]) of follower 2.
Finally, we obtain the equilibrium actions of the two-stage
game--and thus those of the original game--using the findings in the
previous two steps. The equilibrium actions are ([a.sup.*.sub.1],
[b.sup.*.sub.1] ([a.sup.*.sub.1]), [a.sup.*.sub.2] ([a.sup.*.sub.1]),
[b.sup.*.sub.2] ([a.sup.*.sub.1])).
Solving the Subgames Starting at the Second Stage In the second
stage, follower 1 and party 2 play a simultaneous-move game, knowing the
action al chosen by leader 1 in the first stage. To obtain follower
1's reaction function, consider his maximization problem. Knowing
the action [a.sub.1], follower 1 seeks to maximize his payoff over his
action [b.sub.1], taking follower 2's action [b.sub.2] as given:
(A1) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
We assume that for each a t in A l and b2 in B2, maximization
problem (A.1) has a unique interior solution, which is denoted by
(A2) [b.sup.BR.sub.1] ([a.sub.1], [b.sub.2]).
This is follower 1's best response to follower 2's action
[b.sub.2], given the action [a.sub.1] of leader 1. Note that, because
the payoff to follower 1 does not depend directly on leader 2's
action [a.sub.2], it is also follower 1's best response to party
2's pair of actions ([a.sub.2], [b.sub.2]), given the action
[a.sub.1] of leader 1. Now that follower 1's reaction function
shows his best response to every possible action that follower 2 might
choose, it comes from follower 1's best response (A.2). We denote
it by
(A3) [b.sub.1] = [b.sup.BR.sub.1] ([b.sub.2]; [a.sub.1]).
To obtain party 2's reaction functions--specifically, the
reaction function for leader 2 and that for follower 2--we need to
consider two separate but related maximization problems. Working
backward, consider first follower 2's maximization problem. Knowing
the action [a.sub.2] chosen by leader 2, follower 2 seeks to maximize
his payoff over his action [b.sub.2], taking follower 1's action
[b.sub.1] as given:
(A4) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
We assume that for each [a.sub.2] in [A.sub.2] and [b.sub.1] in
[B.sub.1], maximization problem (A.4) has a unique interior solution,
which is denoted by
(A5) [b.sup.BR.sub.2] ([a.sub.2], [b.sub.1]).
This is follower 2's best response to leader 2's action
[a.sub.2] and follower 1's action [b.sub.1].
Next, consider leader 2's maximization problem. Leader 2 seeks
to maximize her payoff over her action [a.sub.2], taking follower
1's action [b.sub.1] as given:
(A6) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Because leader 2 can solve follower 2's maximization problem
(A.4) as well as follower 2 can, leader 2 has perfect foresight about
follower 2's best response to each action [a.sub.2] that she might
take--that is, she knows in advance [b.sup.BR.sub.2] ([a.sub.2],
[b.sub.1]) for each action [a.sub.2]. Thus, leader 2's maximization
problem (A.6) amounts to
(A7) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
We assume that for each [b.sub.1] in [B.sub.1], maximization
problem (A.7) has a unique interior solution, which is denoted by
(A8) [a.sup.BR.sub.2]([b.sub.1]).
This is leader 2's best response to follower 1's action
[b.sub.1].
Follower 2's reaction function shows his best response to
every possible pair of actions that leader 2 and follower 1 might
choose. Thus, it comes from follower 2's best response (A.5):
[b.sub.2] = [b.sup.BR.sub.2] ([a.sub.2], [b.sub.1]). Similarly, leader
2's reaction function shows her best response to every possible
action that follower 1 might choose, and thus it comes from leader
2's best response (A.8): [a.sub.2] = [a.sup.BR.sub.2] ([b.sub.1]).
Therefore, the reaction functions for party 2 are
(A9) [a.sub.2] = [a.sup.BR.sub.2] ([b.sub.1])
and
(A10) [b.sub.2] = [b.sup.BR.sub.2] ([a.sub.2], [b.sub.1]).
Now we obtain the triple vector, ([b.sup.*.sub.1]([a.sub.1]),
([a.sup.*.sub.2]([a.sub.1]), [b.sup.*.sub.2]([a.sub.1])), using the
three reaction functions, (A.3), (A.9), and (A.10). Specifically, we
obtain it by solving the system of three simultaneous equations that
consists of (A.3), (A.9), and (A.10).
The First Stage and the Equilibrium Actions of the Game.
Consider the first stage in which leader 1 chooses her action.
Leader 1 seeks to maximize her payoff over her action [a.sub.1]:
(A11) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Because leader 1 can solve the subgames starting at the second
stage, she knows in advance ([b.sup.*.sub.1]([a.sub.1]),
([a.sup.*.sub.2]([a.sub.1]), [b.sup.*.sub.2]([a.sub.1])) for each action
[a.sub.1] that she might take. Thus, leader 1's maximization
problem (A.11) amounts to
(A12) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
We assume that maximization problem (A.12) has a unique interior
solution which is denoted by [a.sup.*.sub.1].
Finally, substituting [a.sup.*.sub.1] into the triple vector,
([b.sup.*.sub.1] ([a.sub.1]), [a.sup.*.sub.2] ([a.sub.1]),
[b.sup.*.sub.2] ([a.sub.1])), we obtain the equilibrium actions of the
game, ([a.sup.*.sub.1], [b.sup.*.sub.1], ([a.sup.*.sub.1]),
[a.sup.*.sub.2], ([a.sup.*.sub.1]), [b.sup.*.sub.2], ([a.sup.*.sub.1])).
REFERENCES
Appelbaum, E., and E. Katz. "Seeking Rents by Setting Rents:
The Political Economy of Rent Seeking." Economic Journal, 97(387),
1987, 685-99.
Baik, K. H. "Winner-Help-Loser Group Formation in Rent-Seeking
Contests." Economics and Politics, 6(2), 1994, 147-62.
--. Two-Player Asymmetric Contests with Ratio-Form Contest Success
Functions." Economic Inquiry, 42(4), 2004, 679-89.
Baik, K. H., B. R. Dijkstra, S. Lee, and S. Y. Lee. "The
Equivalence of Rent-Seeking Outcomes for Competitive-Share and Strategic
Groups." European Journal of Political Economy, 22(2), 2006,
337-42.
Baik, K. H., and S. Lee. "Strategic Groups and Rent
Dissipation." Economic Inquiry, 39(4), 2001, 672-84.
--. "Collective Rent Seeking When Sharing Rules Are Private
Information." European Journal of Political Economy, 23(3), 2007,
768-76.
Baik, K. H., and J. F. Shogren. "Strategic Behavior in
Contests: Comment." American Economic Review, 82, 1992, 359-62.
--. "Competitive-Share Group Formation in Rent-Seeking
Contests." Public Choice, 83(1-2), 1995, 113-26.
Baye, M. R., and H. C. Hoppe. "The Strategic Equivalence of
Rent-Seeking, Innovation, and Patent-Race Games." Games and
Economic Behavior, 44(2), 2003, 217-26.
Che, Y.-K., and I. Gale. "Rent Dissipation When Rent Seekers
Are Budget Constrained." Public: Choice, 92(1-2), 1997, 109-26.
Davis, D. D., and R. J. Reilly. "Rent-Seeking with
Non-identical Sharing Rules: An Equilibrium Rescued." Public
Choice, 100(1-2), 1999, 31-38.
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.
Fershtman, C., and K. L. Judd. "Equilibrium Incentives in
Oligopoly." American Economic Review, 77(5), 1987, 927-40.
Garfinkel, M. R. "Stable Alliance Formation in Distributional
Conflict." European Journal of Political Economy, 20(4), 2004,
829-52.
Hausken, K. "The Dynamics of Within-Group and Between-Group
Interaction." Journal of Mathematical Economics, 24(7),
1995,655-87.
--. "Production and Conflict Models Versus Rent-Seeking
Models." Public Choice, 123(1-2), 2005, 59-93.
Hillman, A. L., and E. Katz. "Risk-Averse Rent Seekers and the
Social Cost of Monopoly Power." Economic Journal, 94(373), 1984,
104-10.
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.
Inderst. R., H. M. Muller, and K. Warneryd. "Distributional
Conflict in Organizations." European Economic Review, 51(2), 2007,
385-402.
Katz, E., S. Nitzan, and J. Rosenberg. "Rent-Seeking for Pure
Public Goods." Public Choice, 65(1), 1990, 49-60.
Konrad, K. A., and H. Schlesinger. "Risk Aversion in
Rent-Seeking and Rent-Augmenting Games." Economic Journal,
107(445), 1997, 1671-83.
Krueger, A. O. "The Political Economy of the Rent-Seeking
Society." American Economic Review, 64(3), 1974, 291-303.
Lee, S. "Endogenous Sharing Rules in Collective-Group
Rent-Seeking." Public" Choice, 85(1-2), 1995, 31-44.
Leininger, W. "More Efficient Rent-Seeking--A Munchhausen
Solution." Public Choice, 75(1), 1993, 43-62.
Morgan, J. "Sequential Contests." Public Choice,
116(1-2), 2003, 1-18.
Muller, H. M., and K. Warneryd.. "Inside versus Outside
Ownership: A Political Theory of the Firm." Rand Journal of
Economics, 32(3), 2001, 527-41.
Nitzan, S. "Rent-Seeking with Non-identical Sharing
Rules." Public Choice, 71(1-2), 1991a, 43-50.
--. "Collective Rent Dissipation." Economic Journal,
101(409), 1991b, 1522-34.
--. "Modelling Rent-Seeking Contests." European Journal
of Political Economy, 10(1), 1994, 41-60.
Nitzan, S., and K. Ueda. "Prize Sharing in Collective
Contests." Working Paper, Bar-Ilan University, 2007.
Posner, R. A. "The Social Costs of Monopoly and
Regulation." Journal of Political Economy, 83(4), 1975, 807-27.
Sklivas, S. D. "The Strategic Choice of Managerial
Incentives." Rand Journal of Economics, 18(3), 1987, 452-58.
Stein, W. E., and A. Rapoport. "Asymmetric Two-Stage Group
Rent-Seeking: Comparison of Two Contest Structures." Public Choice,
118(1-2), 2004, 151-67.
Tullock, G. "The Welfare Costs of Tariffs, Monopolies, and
Theft." Western Economic Journal, 5(3), 1967, 224-32.
--. "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.
Ueda, K. "Oligopolization in Collective Rent-Seeking."
Social Choice and Welfare, 19(3), 2002, 613-26.
Yildirim, H. "Contests with Multiple Rounds." Games and
Economic Behavior, 51(1), 2005, 213-27.
(1.) Nitzan (1994) provides an excellent survey of the literature
on rent seeking.
(2.) Considering contests in which two players compete for a prize
by choosing their effort levels simultaneously, Dixit (1987) defines the
favorite [the underdog] as the player who has a probability of winning
greater [less] than 1/2 at the Nash equilibrium.
(3.) Balk (1994), Baik and Lee (2001), and Balk et al. (2006) study
collective rent seeking in which the prize is awarded to one of the
players. Alternatively, we can develop a model in which the prize is
awarded to one of the groups; the players in the winning group share the
prize among themselves: the fractional share of player k is determined
by
[[lambda].sub.k] = [theta][x.sub.k] / X + (1 - [theta]) / n,
where [x.sub.k] represents the effort level expended by player k in
the winning group, X = [[summation].sup.n.sub.k=1] [x.sub.k], and the
parameter [theta] is chosen by the players at the beginning of the
contest. This sharing rule specification is used in Nitzan (1991a,
1991b), Baik and Shogren (1995), Hausken (1995), Lee (1995), Davis and
Reilly (1999), Ueda (2002), Baik and Lee (2007), and Nitzan and Ueda
(2007). Note that, in this alternative model, the players in the winning
group need to know how much effort each player expended when they share
the prize, whereas this is not the case with the model under
consideration. Utilizing the results in Baik et al. (2006), one can see
that this alternative model yields exactly the same (main) results.
(4.) This logit-form contest success function is extensively used
in the literature on rent seeking. Examples include Tullock (1980),
Appelbaum and Katz (1987), Hillman and Riley (1989), Hirshleifer (1989),
Katz, Nitzan, and Rosenberg (1990), Nitzan (1991a, 1991b), Leininger
(1993), Balk (1994), Lee (1995), Che and Gale (1997), Hurley and Shogren
(1998), Davis and Reilly (1999), Balk and Lee (2001, 2007), Morgan
(2003), and Stein and Rapoport (2004).
(5.) We obtain the same results with the assumption that
[[sigma].sub.i] > 0. However, the current assumption makes the
analysis simpler.
(6.) Why do such groups exist? Why are such groups formed? Baik and
Lee (2001) provide two reasons. First, depending on the sharing rule,
each player in such a group can share with the other members the risk of
his failure in winning the prize, or can earn more than the prize when
he becomes the winner. Second, the players in such a group can benefit
by achieving strategic commitments through their sharing rule.
Examples of such groups include political parties, R&D joint
ventures among firms, and coalitions among political parties or interest
groups. Some states in the United States provide their colleges and
universities with state funds to match federal awards for research and
research equipment. This gives another example. We can consider each
state as a group in our model, and its colleges and universities as the
players in the group which compete for federal research awards. College
football conferences are yet another example.
(7.) Note that, because the players in each group are identical,
their decision on the sharing rule is unanimous.
(8.) Thus, the game is overall a simultaneous-move game between the
two groups. Furthermore, because the game
has sequential moves, it is a simultaneous-move game with
sequential moves. Baik and Lee (2007) study a general model of the
simultaneous-move game with sequential moves.
(9.) It is straightforward to see that [[pi].sub.ik] in Equation
(1) is strictly concave in [x.sub.ik], and thus the second-order
condition for maximizing Equation (1) is satisfied. Certainly, the
second-order condition is satisfied for every maximization problem in
the paper, although we do not state so explicitly in each case for
concise exposition.
(10.) In the Appendix, we set up and analyze a general model of the
game between two parties in which each party has two sequential moves
and both parties' first moves are not public information.
(11.) In this way of solving the (R, NR) subgame, group 1 is
treated as the strategical leader that chooses its sharing rule before
group 2 does. This may involve assuming that group 2 is not committed to
a sharing rule until group 1 announces its sharing rule publicly.
(12.) Using Equation (13), we see that the players in group 1
expend zero effort when [v.sub.2] ([[sigma].sub.2] - 1/n) [greater than
or equal to] [gamma][v.sub.1][[sigma].sub.1; the players in group 2
expend zero effort when [gamma][v.sub.1] ([[sigma].sub.1] - 1//n)
[greater than or equal to] [v.sub.2][[sigma].sub.2]. Such
"monopolization" in collective rent seeking was first studied
by Ueda (2002). Because monopolization does not occur in the
subgame-pertect equilibrium of the bilateral-release subgame, we omit a
complete description of the symmetric Nash equilibrium of the second
stage for concise exposition.
(13.) In Lemma 4, both [x.sup.BR.sub.1] and [x.sup.BR.sub.2] are
positive only when n/(n + 1) < [alpha][gamma] < (n + 1)/n.
Therefore, Proposition 1 is valid if and only if n/(n + 1) <
[alpha][gamma] < (n + 1)/n.
(14.) Note that [[pi].sup.BR.sub.i] is always smaller than
[[pi].sup.NR.sub.i], smaller than [[pi].sup.1R.sub.i], and smaller than
[[pi].sup.2R.sub.i], for i = 1,2.
(15.) See, for example, Fershtman and Judd (1987) and Sklivas
(1987). Yildirim (2005) studies the following two-player contest. In
period 0, each player decides whether he will release the information
about his period-I effort level to the rival. Then the players announce
their decisions simultaneously. In period 1, knowing which player will
or will not release the information, the players simultaneously choose
their effort levels. Then each player releases the information about his
effort level if he decided to do so in period 0. In period 2, the
players simultaneously choose their effort levels again--that is, each
player adds zero or positive effort to his period-I effort level.
Interestingly, Yildirim finds that both players decide to release the
information in equilibrium.
(16.) Indeed, if [alpha][gamma] > 1, then each player expends a
smaller effort level in the equilibrium involving (NR, R) than in the
equilibrium involving (R, NR); if [alpha][gamma] < 1, then the
opposite holds true. In terms of symbols, if ay > 1, then we have
[x.sup.2R.sub.i] < [x.sup.1R.sub.i] for i = 1,2; if [alpha][gamma]
< 1, then we have [x.sup.1R.sub.i] < [x.sup.2R.sub.i] for i = l,
2.
(17.) Baik and Shogren (1992) and Leininger (1993) study contests
with two asymmetric players in which the players first announce publicly
when they will exert their effort, and then based on this timing, they
choose their effort levels. They find that the underdog always exerts
effort before the favorite does.
(18.) The favorite and underdog choose the same winner's
fractional share if, and only if, n = 2.
(19.) If [alpha][gamma] = 1, then the groups choose the same
winner's fractional share in equilibrium, which is equal to (n +
l)/2n. Note that the equilibrium winner's fractional share depends
only on the size n of the groups; it is greater than a half, but less
than unity. If [alpha][gamma] = 1, then the players in both groups have
the same probability of winning in equilibrium.
(20.) More precisely, [X.sup.2R.sub.1] + [X.sup.2R.sub.2] <
[X.sup.BR.sub.1] + [X.sup.BR.sub.2] in the first part holds when [alpha]
[greater than or equal to] 0.375, and [X.sup.1R.sub.1] +
[X.sup.1R.sub.2] < [X.sup.BR.sub.1] + [X.sup.BR.sub.2] in the second
part holds when 1/[alpha] [greater than or equal to] 0.375. If
[alpha][gamma] = 1, then we have [x.sup.NR.sub.i] = [x.sup.1R.sub.i] =
[x.sup.2R.sub.i] < [x.sup.BR.sub.i] for i = 1,2. This implies that
the total effort level is smaller in the equilibria than in the (R. R)
subgame.
(21.) In the literature on rent seeking, examining the extent of
rent dissipation is one of the main issues because the opportunity costs of resources expended on rent-seeking activities are viewed as social
costs.
Many papers show that less than complete dissipation of the
contested rent occurs. Examples include Tullock (1980), Hillman and
Riley (1989), Baik and Lee (2001), and Baik (2004).
(22.) Note that the (NR, NR) subgame is the case where sharing
rules are private information and the (R, R) subgame is the case where
sharing rules are public information.
(23.) We thank one of the referees for pointing out this possible
commitment problem. In general, a commitment problem refers to a
situation in which players cannot "achieve their goals"
because of their inability to (credibly) commit to their actions.
(24.) It is computationally intractable to show explicitly that
[[pi].sup.2R.sub.1] is greater than or equal to [[pi].sup.BR.sub.1], and
that [[pi].sup.1R.sub.2] is greater than or equal to
[[pi].sup.BR.sub.2].
(25.) We investigated using more than ten different sets of
numerical values, and found that it held true for all the numerical
cases considered.
KYUNG HWAN BAIK and DONGRYUL LEE *
* We are grateful to Susan Feigenbaum, Hanjoon Jung, Eric Rasmusen,
anonymous referees, and seminar participants at Korea University and
Sungkyunkwan University for their helpful comments and suggestions. We
are also grateful to Jongmin Jeon for excellent research assistance. An
earlier version of this paper was presented at the 76th Annual
Conference of the Southern Economic Association, Charleston, SC,
November 2006. Part of this research was conducted while the first
author was a visiting professor at Virginia Polytechnic Institute and
State University. This paper was supported by Faculty Research Fund,
Sungkyunkwan University, 2008.
Baik: Professor, Department of Economics, Sungkyunkwan University,
Seoul 110-745, Korea. Phone 4-82-2-7600432, Fax 4-82-2-760-0946, E-mail
khbaik@skku.edu
Lee: Instructor, Department of Economics, Virginia Polytechnic
Institute and State University, Blacksburg, VA 24061. Phone +
1-540-231-5764, Fax 4-1-540-231-5097, E-mail lee78@vt.edu
TABLE 1
The Strategic Interaction between the Groups
in the First Stage
Group 2
NR R
Group 1 NR [[pi].sup.NR.sub.1], [[pi].sup.2R.sub.1],
[[pi].sup.NR.sub.2] [[pi].sup.2R.sub.2],
R [[pi].sup.1R.sub.1], [[pi].sup.BR.sub.1],
[[pi].sup.1R.sub.2] [[pi].sup.BR.sub.2]
TABLE 2
The Outcomes of the Contest in the Selected
Equilibrium
[alpha][gamma] > 1
Favorite Group 1
First-stage (NR, R)
combination
Strategical Group 2
leader
Winner's 0.5 < [[sigma].sup.2R.sub.1]
fractional [less than or equal to]
shares [[sigma].sup.2R.sub.2] < 1
Probabilities [p.sup.2R.sub.2] < [p.sup.2R.sub.1]
of winning
[alpha][gamma] < 1
Favorite Group 2
First-stage (R, NR)
combination
Strategical Group 1
leader
Winner's 0.5 < [[sigma].sup.1R.sub.2]
fractional [less than or equal to]
shares [[sigma].sup.1R.sub.1] < 1
Probabilities [p.sup.1R.sub.1] < [p.sup.1R.sub.1]
of winning