Classical cooperative theory. II: Value-like concepts (Q4352255)

[For the review of part I see the review Zbl 0917.90297 above.] In some practical conflict situations, we expect game theory to provide a set of possible reasonable outcomes, thus giving the participants the ability to negotiate between them. The notions of core, stable set, etc., correspond to these situations. However, in other practical situations, the participants may completely rely on the arbiter; in such situations, they expect a unique outcome. There are several different approaches to finding the unique outcome. The first approach -- due to Shapley -- originated from games with transferable utility. In this approach, crudely speaking, we award, to each participant, the average benefit that this person brings to a coalition by joining it. The resulting notion of a \textit{Shapley value} has many useful applications and modifications. The second approach -- due to Harsanyi -- originated from games with non-transferable utilities. In these games, it is not easy to describe fairness because, due to non-transferability of utilities, utilities of different participants may correspond to different scales. For every sequence \(\lambda\) of rescaling weights \(\lambda_1,\ldots,\lambda_n\), we can formulate a \textit{\(\lambda\)-fair} solution as a solution in which the (rescaled) utilities of all the players are equal: \(\lambda_1u_1=\lambda_2u_2=\ldots\); we can also formulate a \textit{\(\lambda\)-optimal} solution in which the sum of (rescaled) payoffs is the largest possible: \(\lambda_1u_1+\lambda_2u_2+\ldots\to\max\). Harsanyi proposed to choose \(\lambda\) for which a \(\lambda\)-optimal solution is \(\lambda\)-fair; it turns out that, under reasonable conditions, this solution exists and is unique. Several other solution concepts are also overviewed in this paper. For the entire collection see [Zbl 0863.00056].