Reduction of Economic Inequality in Combinatorial Domains
Ulle Endriss
Abstract:
Criteria for measuring economic inequality, such as the Lorenz curve
and the Gini index, are widely used in the social sciences but have
hardly been explored in Multiagent Systems, even though the
significance of other concepts from fair division is widely accepted
in the field. In a departure from the standard model used in
Economics, we apply inequality criteria to allocation problems with
indivisible goods, i.e., to the kind of problem typically analysed in
Multiagent Systems. This gives rise to the combinatorial optimisation
problem of computing an allocation that reduces inequality with
respect to an initial allocation (and the closely related problem of
minimising inequality), for a chosen inequality measure. We define
this problem, we discuss the computational complexity of various
aspects of it, and we formulate a generic approach to designing
modular algorithms for solving it using integer programming.