NotesOn the nonemptiness of the α-core of discontinuous games: Transferable and nontransferable utilities☆
Introduction
Two major solution concepts for normal form games are the Nash equilibrium and the core. A Nash equilibrium is a noncooperative solution in which the joint interest of groups of players is not explicitly considered, whereas the core is a cooperative solution involving the behavior of coalitions of players. Motivated by economic problems that are suitably modeled by games with discontinuous payoff functions, Nash's existence result has been considerably extended following the seminal works of Dasgupta and Maskin (1986) and Reny (1999).1 The existence of a cooperative solution for such games has yet to be provided. This note fulfills this objective.
An action profile belongs to the core of a game if no group of players has an incentive to form a coalition in which each of its members are made better-off, i.e. the action profile cannot be blocked by any coalition. In a (normal form) game, the actions of the complementary coalition affect the payoff of the members of a coalition, and therefore the definition of “blocking” hinges crucially on what the complementary coalition does. Among various blocking concepts defined in the literature, the α-core due to Aumann (1961) has attracted a significant attention.2 In this paper we study the existence of this cooperative solution for games with possibly discontinuous payoff functions without transferable utilities (α-core) and with transferable utilities (-core).
An action profile is in the α-core of a game if no coalition has an alternative action which makes all of its members better off, independently of the actions of the complementary coalition. Hence it is a pessimistic solution concept regarding the actions of the complementary coalition. And a pair of an action profile and a payoff profile is in the -core of a game if the action profile maximizes the grand coalition's aggregate payoff, and no coalition has an alternative action which guarantees a higher aggregate payoff, independently of the actions of the complementary coalition. Therefore, the α-core allows the coordination of the actions among the members of the coalitions, whereas the -core allows payoff transfers within the coalitions in addition to the coordination of actions. Scarf (1971) and Zhao (1999) proved the following theorems.
Theorem Scarf Let be a game such that for each player i, is a nonempty, compact and convex subset of a finite dimensional Euclidean space, is a quasiconcave and continuous function on .
Then G has a nonempty α-core.
Theorem Zhao Let be a game such that for each player i, is a nonempty, compact and convex subset of a finite dimensional Euclidean space, is a concave and continuous function on , G is weakly separable.
Then G has a nonempty -core.
In this paper we generalize these results to games where the continuity assumptions are weakened. And in line with the recent literature on discontinuous games pioneered by Reny (1999), the notions of coalitionally C-secure, coalitionally -secure and coalitionally -secure games are presented, and the existence of an imputation in the α-core and -core are shown for them.
The paper is organized as follows. Section 2 defines the basic concepts and states the results, Section 3 illustrates examples, Section 4 provides proofs of the results, and Section 5 concludes.
Section snippets
The model and the results
A (normal form) game is a list where
- (i)
is the finite set of players,
- (ii)
is the nonempty set of actions of player ,
- (iii)
is the utility function of player defined on .
In his pioneering work, Reny (1999) proved the existence of a Nash equilibrium of
Examples
The first example illustrates a game which does not have a (pure strategy) Nash equilibrium, but has a nonempty α-core and -core. The game satisfies all the assumptions of Proposition 1, Proposition 2, particularly coalitional C-security and coalitional -security. However, it neither satisfies C-security nor coalitional -security.
Example 1 Consider the following Bertrand duopoly game. Each firm's action set is . The market demand function is defined as And the
Proofs of the results
The proofs of Theorem 1, Theorem 2, Theorem 3 require a delicate construction. In order to give a general overview of the proofs, we first provide a heuristic outline of the proof of Theorem 1. Whereas the details of the proofs of Theorem 2, Theorem 3 are different, the construction in the proof is quite similar. Our proof of Theorem 1 is by contradiction. We assume the α-core of the game is empty. By using the compactness of the action sets and the coalitional C-security of the game, we obtain
Concluding remarks
This paper provides sufficient conditions for the nonemptiness of the TU and NTU α-cores of games with possibly discontinuous payoff functions. We end this paper with three remarks. First, although the α-core is widely applied cooperative solution concept for normal form games, a number of different solution concepts, such as β-core, strong equilibrium and hybrid solution of Zhao (1999) are also of interest for analyzing specific problems. It will be of interest to discuss the existence of such
References (19)
Bertrand without fudge
Econ. Lett.
(2003)A generalization of Scarf's theorem: an α-core existence theorem without transitivity or completeness
J. Econ. Theory
(1992)- et al.
Equilibrium binding agreements
J. Econ. Theory
(1997) On the existence of a cooperative solution for a general class of n-person games
J. Econ. Theory
(1971)The core of a cooperative game without side payments
Trans. Am. Math. Soc.
(1961)- et al.
A note on the equilibrium existence problem in discontinuous games
Econometrica
(2013) The theory of the core in an n-person game
Vestn. Leningr. Univ.
(1962)Symposium on: existence of Nash equilibria in discontinuous games
Econ. Theory
(2011)- et al.
The existence of equilibrium in discontinuous economic games, I: theory, II: applications
Rev. Econ. Stud.
(1986)
Cited by (33)
The core in an N-firm dynamic Cournot oligopoly
2024, Mathematical Social SciencesExistence and well-posedness of the α-core for generalized fuzzy games
2023, Fuzzy Sets and SystemsCitation Excerpt :After that, based on the existence of α-core by Scarf and Border, Kajii [7] not only proved the existence of α-core for n-person noncooperative games, but also extended this existence result to generalized games. Recently, Uyanik [8] provided sufficient conditions for the existence result of NTU and TU α-core in normal form games with discontinuous and unbounded payoff function. Scalzo [9] obtained the sufficient and necessary conditions of the existence of α-core with the assumption of disordered, discontinuous preferences and infinite dimensional strategy space.
Cooperative concurrent games
2023, Artificial IntelligenceCitation Excerpt :In this section, we present some conclusions, briefly discuss related work, discuss some issues related to the core, and speculate about how to implement our approach using model checking techniques. Coalition formation with externalities has been studied in the cooperative game-theory literature [16,39,40]. These works considered several possible formulations of the core.
On the strong hybrid solution of an n-person game
2022, Mathematical Social SciencesA weak α-core existence theorem of generalized games with infinitely many players and pseudo-utilities
2022, Mathematical Social Sciences
- ☆
The results reported here constitute part of the author's Ph.D. dissertation being written under the supervision of Professors Hülya Eraslan and M. Ali Khan. I am grateful to them and to Richard P. McLean for their careful reading of the preliminary drafts. I also thank Ying Chen, Idione Meneghel, Burçin Kısacıkoğlu, Sezer Yaşar and the seminar participants at the University of Queensland and at the Sunday Theory Workshops at Johns Hopkins University for stimulating discussion. Finally, I am grateful to an anonymous referee of this journal for his/her report and for bringing Zhao (1999) to my attention.