Cores and large cores when population varies (Q752022)

The author examines the problem of population monotonicity (when population increases, all agents should have a nonnegative share in the external costs or benefits imposed by the increase) in the context of the core. The basic model is of a large potential population N. Then monotonicity is defined with respect to population increases from S to T where \(S\subset T\subset N.\) Let (N,v) be a game in characteristic function form. A generalized allocation is a vector \(x=(x^ S)_{S\subset N}\) such that \(x^ S\in R^ S\) and \(\sum_{i\in S}x^ S_ i=v(S)\). A generalized allocation x is in the monotonic core if \(i\in T\subset S\) implies \(x^ T_ i\leq x^ S_ i\). This implies that \(x^ S\) is in the core of \((S,v^ S)\), \(v^ S\) being the restriction of v to \(\{\) \(T\subset S\}\). The author shows that the convex games are strictly contained in the set of games with a monotonic core, which is strictly contained in the set of balanced games. A generalized aspiration \(y=(y^ S)\) of (N,v) satisfies \(\sum_{i\in T}y^ S_ i\geq v^ S(T)\) for all \(T\subset S\subset N\). A generalized aspiration x which is a generalized allocation clearly induces a core allocation \(x^ S\) in each subgame \((S,v^ S)\), (N,v) has a totally large core if every generalized aspiration y has a generalized allocation x with \(x\leq y\) and \(x^ S\) a core allocation of \((S,v^ S)\) for every S. The first main result is Theorem 1: (N,v) is convex iff it has a totally large core. A monotonic aspiration is a generalized aspiration y such that \(i\in T\subset S\) implies \(y^ T_ i\leq y^ S_ i\). The second main result is Theorem 2: In a convex game a monotonic aspiration satisfying \(y^ S_ i=y_ i^{S'}\) whenever \(i\in S\), \(i\in S'\), and \(| S| =| S'|\) will bound above some monotonic core allocation. This result is quite a bit weaker than existence of a ``large monotonic core'', but is of use when some uniformity, anonymity, or egalitarian property is desired.