Search

Why do people give when asked, but prefer not to be asked, and even take when possible? We introduce a novel analytical framework that allows us to express context dependence and narrow bracketing axiomatically. We then derive the utility representation of distributive preferences additionally obeying standard axioms such as separability and scaling invariance. Such pref- erences admit a generalized prospect-theoretical utility representation reminiscent of fairness- based altruism. As in prospect theory, the underlying preferences are reference dependent and non-convex, which directly predicts the previously irreconcilable empirical evidence on giving, sorting, and taking. We test the model quantitatively on data from seminal experiments and observe significantly improved fit in relation to existing models, both in-sample and out-of- sample.

Reanalyzing 12 experiments on the repeated prisoner's dilemma (PD), we robustly observe three distinct subject types: defectors, cautious cooperators and strong cooperators. The strategies used by these types are surprisingly stable across experiments and uncorrelated with treatment parameters, but their population shares are highly correlated with treatment parameters. As the discount factor increases, the shares of defectors decrease and the relative shares of strong cooperators increase. Structurally analyzing behavior, we next find that subjects have limited foresight and assign values to all states of the supergame, which relate to the original stage-game payoffs in a manner compatible with inequity aversion. This induces the structure of coordination games and approximately explains the strategies played using Schelling's focal points: after (c,c) subjects play according to the coordination game's cooperative equilibrium, after (d,d) they play according to its defective equilibrium, and after (c,d) or (d, c) they play according to its mixed equilibrium.

Multinomial logit is the canonical model of discrete choice but widely criticized for requiring functional form assumptions as foundation. The present paper shows that logit is behaviorally founded without such assumptions. Logit\'s functional form obtains if relative choice probabilities are independent of irrelevant alternatives and invariant to utility translation, to relabeling options (presentation independence), and to changing utilities of third options (context independence). Reviewing behavioral evidence, presentation and context independence seem to be violated in typical experiments, though not IIA and translation invariance. Relaxing context independence yields contextual logit (Wilcox, 2011), relaxing presentation independence allows to capture \"focality\" of options.

Why do people give when asked, but prefer not to be asked, and even take when possible? We show that standard behavioral axioms including separability, narrow bracketing, and scaling invariance predict these seemingly inconsistent observations. Specifically, these axioms imply that interdependence of preferences (\"altruism\") results from concerns for the welfare of others, as opposed to their mere payoffs, where individual welfares are captured by the reference-dependent value functions known from prospect theory. The resulting preferences are non-convex, which captures giving, sorting, and taking directly. Re-analyzing choices of 981 subjects in 83 treatments covering many variants of dictator games, we find that individual reference points are distributed consistently across studies, allowing us to classify subjects as either non-givers, altruistic givers, or social pressure givers and use welfare-based altruism to reliably predict giving, sorting, and taking across experiments.

When do we cooperate and why? This question concerns one of the most persistent divides between \"theory and practice\", between predictions from game theory and results from experimental studies. For about 15 years, theoretical analyses predict completely-mixed \"behavior\" strategies, i.e. strategic randomization rendering \"when\" and \"why\" questions largely moot, while experimental analyses seem to consistently identify pure strategies, suggesting long-run interactions are deterministic. Reanalyzing 145,000 decisions from infinitely repeated prisoner\'s dilemma experiments, and using data-mining techniques giving pure strategies the best possible chance, we conclude that subjects play semi-grim behavior strategies similar to those predicted by theory.

Existing theoretical and experimental studies have established that unanimity is a poor decision rule for promoting information aggregation. Despite this, unanimity is frequently used in committees making decisions on behalf of society. This paper shows that when committee members are exposed to "idiosyncratic" payoffs that condition on their individual vote, unanimity can facilitate truthful communication and optimal information aggregation. Theoretically, we show that since agents" votes are not always pivotal, majority rule suffers from a free-rider problem. Unanimity mitigates free-riding since responsibility for the committee's decision is equally distributed across all agents. We test our predictions in a controlled laboratory experiment. As predicted, if unanimity is required, subjects are more truthful, respond more to others' messages, and are ultimately more likely to make the optimal decision. Idiosyncratic payoffs such as a moral bias thus present a rationale for the widespread use of unanimous voting.

Experimenters have to make theoretically irrelevant decisions concerning user interfaces and ordering or labeling of options. Such presentation decisions affect behavior and cause results to appear contradictory across experiments, obstructing utility estimation and policy recommendations. The present paper derives a model of choice allowing analysts to control for both presentation effects and stochastic errors in econometric analyses. I test the model in a comprehensive re-analysis of dictator game experiments. Controlling for presentation effects, preference estimates are consistent across experiments and predictive out-of-sample, highlighting the fundamental role of presentation for choice, and this notwithstanding the possibility of reliable estimation and prediction.

The common use of majority rule in group decision making is puzzling. In theory, it inequitably favors the proposer, and paradoxically, it disadvantages voters further if they are inequity averse. In practice, however, outcomes are equitable. The present paper analyzes data from a novel experimental design to identify the underlying social preferences. Our experiment compares one-shot and indefinite horizon versions of random-proposer majority bargaining (the Baron-Ferejohn game) which allow us to disentangle behaviors compatible with altruism, inequity aversion, and reference dependent altruism. Most subjects are classified as reference-dependent altruists, around 10% are inequity averse. Subjects are egoistic when their payoff is below their reference point, they become efficiency concerned when satisfied, and the reference point is either the ex ante expectation or the opponent\'s payoff. Finally, we successfully test RDA out-of-sample on a number of distribution and bargaining games from three seminal social preference experiments.

In view of behavioral patterns left unorganized by current social preference theories, we propose a theory of reference dependent altruism (RDA). With RDA, one's degree of altruism increases at reference points. It induces equity and efficiency effects that are conditional on whether or not payoffs meet reference points. We verify the theory first by experimentally analyzing majority bargaining, where observed behavior contradicts existing theories but confirms RDA. Using parameter estimates from majority bargaining, we then make out-of-sample predictions for Charness-Rabin, Engelmann-Strobel, and Bolton-Ockenfels games. RDA organizes these seemingly disparate games out-of-sample, which validates our hypothesis that pro-social behavior primarily relates to reference points.

We show and explain how generosity beyond that explainable by social preferences can manifest in bargaining. We analyze an ultimata game with two parties vying to coalesce with a randomly chosen proposer. They simultaneously demand shares of the surplus. The proposer must then make an offer that meets at least one demand, or else the game either continues with a new round or breaks down with all earning zero. Self-interest, altruism, and inequity aversion univocally predict miniscule demands due to inter-party competition; proposers thus obtain the lion's share. We experimentally observe that proposers coalesce with the less demanding party by strategically matching demands, like ultimatum bargaining, but also give non-strategically to the other party, like dictator giving. The observations are incompatible with concave utilities, as implied by social preferences, but are compatible with reference dependent preferences.

This paper proposes two generalization of the core and evaluates them on experimental data of assignment games (workers and firms negotiate wages and matching). The generalizations proposed allow for social utility components (e.g. altruism) and random utility components (e.g. logistic perturbations). These generalizations are well-established in analyses of non-cooperative games, and they prove to be both descriptive and predictive in the assignment games analyzed here. The "logit core" allows us to define a "stochastically more stable" relation on the outcome set, which has intuitive implications, and it fits better than alternative approaches such as random behavior cores and regression modeling.

In a menu auction, players submit bids for all choices the auctioneer A can make, and A then makes the choice that maximizes the sum of bids. In a binomial menu auction (BMA), players submit acceptance sets (indicating which choices they would support), and A chooses the option that maximizes his utility subject to acceptance of the respective players. Monetary transfers may be implicit, but players may also bid by offering "favors" and the like. BMAs provide a unified representation of both monetary and non-monetary bidding, which I apply to model government formation. First, I analyze general BMAs, characterize the solution under complete information and establish outcome uniqueness (for both, sealed bid and Dutch formats). Second, in case monetary transfers are possible, BMAs are shown to implement VCG mechanisms. Finally, in case transfers are impossible, BMAs extend the model of proto-coalition bargaining and are specifically applied to government formation.

Are "generous" bargaining offers made out of fairness or in fear of rejection? We disentangle risk and social references by analyzing experimental behavior in three majority bargaining games: (1) a random-proposer game with infinite time horizon; 2) a one round proposer game with disagreement payoffs equal to the infinite horizon continuation payoffs; and, (3) a demand commitment game. Inequity aversion predicts very differently across these games, but risk aversion does not. Observed strategies violate neither stationarity nor truncation consistency. This allows us to use structural models of bargaining behavior to estimate the latent type shares of subjects with CES, inequity averse, and Prospect theoretic preferences. The Prospect theoretic, i.e. reference-dependent, model of utility explains the observations far better than any mixture of alternative models.

The paper analyzes econometric models of altruistic giving in dictator and public goods games. Using existing data sets, I evaluate internal and external validity of "atheoretic" regression models as well as structural models of random behavior, random coefficients, and random utility, controlling for subject heterogeneity by finite mixture modeling. In dictator games, atheoretic regression lacks external validity, while random coefficient models and random utility models offer high degrees of both internal and external validity. In public goods games, regression works comparably well, being bettered only by random utility models. Overall, the ordered GEV model of random utility is most appropriate to describe choices in the considered games.