Other-Regarding Preferences and Consequentialism

Thumbnail Image
Degree type
Doctor of Philosophy (PhD)
Graduate group
Other-Regarding Preferences
Stochastic Dominance
Procedural Concerns
Expressive Voting
Economic Theory
Grant number
Copyright date
Related resources

This dissertation addresses a basic difficulty in accommodating other-regarding preferences within existing models of decision making. Decision makers with such preferences may violate the property of stochastic dominance that is shared by both expected utility and almost any model of non-expected utility. At its core, stochastic dominance requires a decision maker's behavior to conform to a basic form of consequentialism, namely, that her ranking of outcomes should be independent of the stochastic process that generates these outcomes. On the other hand, decision makers with other-regarding preferences may show a concern for procedures; that is they may care not just about what the outcomes of others are but also about how these outcomes are generated and therefore their ranking of outcomes may be intrinsically dependent on the outcome-generating process. We provide theoretical foundations for a new representation of other-regarding preferences that accommodates concerns for procedure and possible violations of stochastic dominance. Our axioms provide a sharp characterization of how a decision maker's ranking of outcomes depends on the procedure by expressing `payoffs' as a weighted average of her concerns for outcomes, and her concerns for procedure. The weight used in evaluating this weighted average, which we call the procedural weight, is uniquely determined and quantifies the relative importance of procedural concerns. In the special case in which procedural concerns are absent our baseline decision model reduces to expected utility, and our most parsimonious representation is one parameter richer than that model. We use our decision model to provide an expressive theory of voting.

Andrew Postlewaite
Date of degree
Date Range for Data Collection (Start Date)
Date Range for Data Collection (End Date)
Digital Object Identifier
Series name and number
Volume number
Issue number
Publisher DOI
Journal Issue
Recommended citation