Replication invariance of bargaining solutions (Q1077341)

The author considers a bargaining problem, i.e. a problem of choosing a single output from the set S of possible outputs. An output is characterized by a sequence \(x_ 1,...,x_ n\) of utilities obtained by each of the n players. Without any loss of generality we can assume that a status quo point is \(x_ i=0\). The famous Nash solution consists in choosing an output with \(\prod x_ i\to \max\). One of the drawbacks of the Nash solution was revealed by Kalai, who noted that if we use the Nash solution then a player can often obtain more if he ficticiously subdivides his interests between two formally independent agents (this procedure is called replication). Kalai suggested a new solution, where the output for the i-th player is proportional to max \(x_ i\), where the maximum is taken over the whole set S, and showed that this solution is replication invariant. The author of the paper under review shows that Kalai's formalization of the notion ''replication'' is adequate only for state-controlled economies because of too many restrictions on the agents' behaviour otherwise. In case we use a more natural formalization of this notion, namely a formalization in which \(x_ i\) in the original game can be arbitrarily subdivided between agents of the i-th player, then Kalai's solution is not replication invariant. Several analogies between the Nash and Kalai solutions are proved.