Research

Abstract

This paper investigates the impact of reciprocal preferences and countries’ fairness-based expectations of each other in the context of international agreements aimed at providing global public goods. Reciprocal countries reward kindness (positive reciprocity) while retaliating against unkind behavior (negative reciprocity). We introduce a model where countries decide not only whether to participate on a coalition but also determine the extent of their effort. We demonstrate that reciprocal preferences can exert both positive and negative effects on the effort of both signatories and non-signatories of a treaty, depending on their expectations of others. In the non-cooperative game, there are three possible Nash equilibria, which depends on the degree of reciprocal concerns and expectations: no contribution, full contribution by all, and an interior solution. In the coalition formation game, if countries have limited yet sufficiently strong reciprocal concerns and low expectations, the grand coalition – where all countries sign the treaty (and exert efficient effort levels) – becomes stable. Interestingly, we find that signatories positively influence non-signatories’ efforts. Anticipating this response, signatories have an additional strategic incentive to increase their efforts, thereby enhancing the overall level of cooperation.

Abstract

Rational inattention models treat information costs as structural features of decision makers, yet the same information choice can often be represented either as an experiment or as the posterior beliefs induced by that experiment. We design a laboratory experiment to test whether the information-cost structure revealed from choice is consistent across these theoretically equivalent representations. Subjects choose from the same feasible set of information levels either by selecting a posterior-accuracy level or by selecting the corresponding experiment-based sampling fraction. We vary both prior beliefs and the reward from a correct decision and evaluate behavior within each treatment using Prior-Invariant (PI) and Uniformly Posterior-Separable (UPS) cost benchmarks. Importantly, neither treatment is assigned to a particular cost class ex-ante. In a second section of the experiment, subjects also choose the representation under which to make an incentivized information decision, allowing us to examine whether the treatment differences persist when the representation is selected by the decision maker rather than externally assigned. 

Information acquisition differs systematically across treatments even though the underlying states, incentives, and feasible information choices are held fixed. A substantial share of subjects exhibit choice patterns that are not fully consistent with either benchmark, and many subjects change the cost class with which their behavior is more consistent across treatments. Aggregate comparative statics similarly resemble different information-cost structures under the two representations. Because PI and UPS are theoretically distinct and generically disjoint under standard conditions, these changes cannot be dismissed as an innocuous reparameterization of a common cost function. The results demonstrate that theoretically equivalent representations of information need not be behaviorally equivalent and raise concerns about transporting revealed information costs across elicitation environments without first establishing the consistency of their representations.

Individuals may be uncertain about their own future preferences, which can generate a demand for flexible menus. At the same time, the presence of a tempting option, especially when accompanied by a recommendation, may induce a preference for commitment to a smaller menu. This paper studies menu choice in an environment where these two forces arise simultaneously. In a lab experiment, I elicit subjects' ex-ante menu preferences and observe their ex-post random choices from the assigned menu to examine the relationship between menu demand and stochastic choice behavior. More specifically, the experiment identifies the extent to which ex-post randomized choice is driven by preference for flexibility or by temptation-induced preference for commitment, and whether these motives can coexist within the same individual. The results contribute to the literature on menu choice by showing how flexibility and commitment jointly shape random choice behavior when future preferences are uncertain and temptation is present.