Cartels via the modiclus. (Q1421474)

The authors study cooperative `linear production' games, where disjoint subsets of players are endowed with the essential factors which are assembled for production; each subset/coalition owns a given amount of one factor, its members possessing fractions of it. The prototype game in this class is the `gloves game' where two coalitions own respectively right and left gloves only and glove pairs must be produced. Classical `equilibrium' allocations like the core assign most of the total profit/value to coalitions owning factors in short supply; in the gloves game for instance, if there are \(N\) left and \(N+1\) right gloves around, owners of right gloves (the long side of the market) get nothing at all however large \(N\) may be. This is rather unnatural, and the second author has previously developed a solution concept, the Modiclus, which takes into account not only what a coalition can attain but also what it can prevent the complementary coalition to achieve. The main novelty with this concept is that if the long side is `not too long', then it gets positive profits in the solution. Modiclus stands for modified nucleoulus, and the concept is easily understood starting from the latter as follows. Given a game \(v\) on player set \(I\), the dual game \(v^\star\) is defined as \(v^\star(S)=v(I)-v(S^c)\) (\(S\)'s power in \(v^\star\) comes from leaving the complementary coalition \(S^c\) with little power in \(v\)). Now define a game with player set given by two copies of \(I\) thus: for coalitions \(S,T\) in the two copies let \(\widetilde{v}(S\cup T)=\max \{v(S)+v^\star(T),v^\star(S)+v(T)\}\). Notice in particular that for \(S=T\) the formula gives the sum of a coalition's `constructive' and `preventive' power. Then the modiclus of the game \(v\) is just the projection on \(I\) of the nucleolus of this derived game \(\widetilde{v}\). The present paper unveils the structure of the modiclus of linear production games, spelling out in detail both the share of total profits allocated to the various coalitions and the internal distribution of each coalition's allotment among its members. We next summarize this structure as described in the main results, using the terms `short side of the market' to mean the set of coalitions owning scarce factors and `long side' for the other coalitions (of course expressions in quotations here and below translate into formal statements in the paper): 1 (Shares of total profit). If the short side is `very short', the long side gets nothing (as in the core), and if the game is large enough the modiclus coincides with the nucleolus; on the other hand if short side is `not too short' (`strong long side'), all coalitions get the same share of total profits. 2 (Distributions within coalitions). In the case of strong long side, provided the game is large enough, the players in long-side coalitions get a payoff proportional to their endowment, the proportionality coefficients being smaller the more the owned factor is in excess supply. Bargaining among players in short-side coalitions on the other hand takes place on the basis of claims players advance according to their potential contributions to external coalitions; these claims add up to more than what they can split (their `estate'), and the modiclus organizes the distribution of the estate according to the `Contested-Garment' solution (devised in the Talmud for a bankruptcy problem analysed as a game by \textit{R. J. Aumann} and \textit{G. Maschler} [J. Econ. Theory 36, 195--213 (1985; Zbl 0578.90100)].