This paper axiomatically characterizes the rule of aggregating individuals’ ambiguous beliefs into the society’s ambiguous belief. We consider a family of social rankings over acts which are indexed by pairs of utility profiles and ambiguous belief profiles, and impose variable-profile axioms which express how society should respond to changes in belief profiles and changes in how beliefs disagree. Our characterization allows two kinds of collective ambiguity aversion, one arising due to individuals’ ambiguity aversion and the other arising due to disagreements in beliefs.
This paper introduces the notion of 𝒢 -rules, which extend the Pareto rule when populations are endogenous. Societal evaluations of lives are required. 𝒢 -rules compare pairs of allocations whenever evaluating the gains of some agents—utility gains or benefits of (not) being born—against the losses of others is not required. 𝒢 -efficiency naturally arises. The 𝒜 and 𝒫 criteria of Golosov et al. (2007) are special cases. Transitive 𝒢 -rules admit an axiomatic foundation based on normatively appealing properties and can be conveniently represented with 𝒢 -nondecreasing critical levels. Many public policies that distort fertility decisions are 𝒢 -efficient because they benefit children who are (not) born as a result of these policies.
Largesse design is a positive theory of what happens when players in a game, who have a limited willingness to sacrifice for the common good, can coordinate on a welfare-optimal solution. This theory provides an alternative to psychological theories in explaining why players engage in costly punishments and rewards. It also says that players will engage in other low-cost methods of encouraging pro-social behavior, for example, through reputation effects. In this paper, I study the properties and comparative statics of the largesse design problem. Despite a lack of concavity in the maximization problem, solutions are unique and continuous except on a lower-dimensional bifurcation set. In examples, I study these bifurcations, in which a small change in the parameters can lead to a large change in behavior or welfare.
I study a model with repeated electoral competition between two parties. Because there is incumbency advantage, the future electoral prospects of a party suffer when it loses an election and strategic voters have an incentive to support their ideologically preferred party even when they dislike its candidate. As a consequence, they vote in a more partisan way than sincere voters with identical preferences. While this form of strategic voting cannot be the only source of an incumbency advantage, it amplifies it and increases the probability that the party whose victory maximizes welfare loses the election. A perceived polarization of the electorate can be caused by voters with unaltered preferences learning to cast their votes in a more sophisticated way.
Classical utility theory provides powerful tools for representing preferences over alternatives, but offers little guidance when the objects of evaluation are preference relations themselves. Such higher-order evaluations are essential for understanding differences of opinion, since the significance of a disagreement depends not only on how two rankings differ but also on how individuals evaluate those differences. This paper develops an axiomatic theory of utility representation for such higher-order evaluations, deriving second-order evaluations of preference relations from observed first-order preferences and providing a foundation for measuring differences of opinion. We introduce behaviorally plausible axioms—independence, disagreement aversion, and symmetry—and show that they characterize a class of hyperpreference relations admitting additively separable utility representations. The resulting hyperutility functions extend utility analysis from preferences over alternatives to preferences over preference relations and allow different types of disagreement to carry distinct utility consequences. We apply this framework to the study of aggregation mechanisms, deriving necessary and sufficient conditions on hyperutility parameters that induce truthful preference reporting and guarantee local efficiency. Our results yield new insights into strategy-proof mechanism design under deep preference heterogeneity and clarify when differences of opinion can be meaningfully and reliably measured.
