Search

Observed individual behavior in the presence of ambiguity shows insufficient responsiveness to changes in subjective likelihoods. Despite being integral to theoretical models and relevant in many domains, evidence on the causes and determining factors of such likelihood insensitive behavior is scarce. This paper investigates the role of beliefs in the form of ambiguity perception – the extent to which a decision-maker has difficulties assigning a single probability to each possible event – as a potential determinant. Using an experiment, I elicit measures of ambiguity perception and likelihood insensitivity and exogenously vary the level of perceived ambiguity. The results provide strong support for a perception-based explanation of likelihood insensitivity. The two measures are highly correlated at the individual level, and exogenously increasing ambiguity perception increases insensitivity, suggesting a causal relationship. In contrast, ambiguity perception is unrelated to ambiguity aversion – the extent to which a decision-maker dislikes the presence of ambiguity. [ABSTRACT FROM AUTHOR]

Although much of the theoretical and applied literature involving decision under ambiguity works under the assumption of uncertainty aversion, experimental evidence suggests that it is not a universal behavioural trait. This paper introduces and axiomatises the family of α -UA (for α -Uncertainty Attitude) preferences: a simple extension of uncertainty averse preferences with a Hurwicz-style mixing coefficient, so as to admit a richer range of uncertainty attitudes. The parameters of the model are uniquely identified in our characterisation. It provides, in the Hurwicz α -maxmin EU special case, a new resolution of a long-standing identification problem. It also yields novel models, including extensions of variational and multiplier preferences. Comparative statics support the interpretation of the mixing coefficient as an index of imprecision aversion. In a standard portfolio problem, the model yields the intuitive relationship between imprecision aversion and investment in an uncertain asset: as the former increases, the latter decreases. [ABSTRACT FROM AUTHOR]

This paper provides a micro-foundation for approximate incentive compatibility using ambiguity aversion. In particular, we propose a novel notion of approximate interim incentive compatibility, approximate local incentive compatibility , and establish an equivalence between approximate local incentive compatibility in a Bayesian environment and exact interim incentive compatibility in the presence of a small degree of ambiguity. We then apply our result to the implementation of efficient allocations. In particular, we identify two economic settings—including ones in which approximately efficient allocations are implementable and ones in which agents are informationally small—in which efficient allocations are approximately locally implementable when agents are Bayesian. Applying our result to those settings, we conclude that efficient allocations are exactly implementable when agents perceive a small degree of ambiguity. [ABSTRACT FROM AUTHOR]

Ambiguous language is ubiquitous and often deliberate. Recent theoretical work (Bose and Renou, 2014 ; Kellner and Le Quement, 2018 ; Beauchêne et al., 2019) has shown how language ambiguation can improve outcomes by mitigating conflict of interest. Our experiment finds a significant effect of language ambiguation on subjects who are proficient at Bayesian updating. For ambiguity averse subjects within this population, a significant part of this effect operates via the channel of subjects' desire to reduce ambiguity. For both ambiguity averse and neutral subjects within this population, an additional behavioral channel is also present. [ABSTRACT FROM AUTHOR]

Semi-consistent conditional preferences are inconsistent enough for different ambiguity attitudes to manifest themselves in different behavior and consistent enough for information to be generically valuable. To simultaneously achieve these two desiderata I assume exactly one type of dynamic inconsistency: agents do not update their preferences upon learning independent randomization outcomes. [ABSTRACT FROM AUTHOR]

This paper explores multi-stage incomplete information games with common ambiguous information about states or types and ambiguity averse players. We characterize a belief formation process that allows players to take their knowledge about the structure of the game into account. This process leads to subjective rectangular ex-ante belief sets for all players. We show that given these sets of beliefs, players behave dynamically consistent. Therefore, using our belief formation process, we can extend the concept of sequential equilibria to multi-stage ambiguous incomplete information games. Furthermore, we characterize assumption under which sequential rationality implies rectangularity. [ABSTRACT FROM AUTHOR]

In the context of normal-form games with complete information, we introduce a notion of correlated equilibrium that allows partial delegation to a mediator and ambiguity in the correlation device. Without ambiguity, the sets of equilibrium action distributions are equivalent to those for coarse correlated equilibrium (Moulin and Vial, 1978). With correlation devices that incorporate ambiguity, any action distribution that Pareto dominates a coarse correlated equilibrium or a correlated equilibrium (Aumann, 1974), can be approximated with an arbitrary degree of precision using the proposed equilibrium notion. These approximations are attained in one-shot, static strategic interactions, and do not require repeated play. We also analyze such equilibria when the set of feasible posteriors is exogenously constrained, which yields, as a special case, a definition and characterization of an "ambiguous correlated equilibrium" that does not require delegation to the mediator. [ABSTRACT FROM AUTHOR]

Outsourcing of research is commonly observed in knowledge-intensive industries such as biotech. We model innovation as an ambiguous stochastic process with dynamically consistent updating. We assume that the research labs are less ambiguity averse than the commercial firms. We characterize the optimal sequence of short-term contracts governing innovation, and show how it facilitates ambiguity-sharing. This ambiguity sharing agreement mitigates the dynamic moral hazard problem, resulting in monotonically decreasing investment and preventing equilibrium delay. Compared to an ambiguity-neutral policymaker's benchmark, the research alliances stop experimenting too early, and may liquidate the project even after being patented. The problems are structural, redesigned patent laws cannot solve both of these problems. [ABSTRACT FROM AUTHOR]

This paper studies a multi-state binary choice experiment in which in each state, one alternative has well understood consequences whereas the other alternative has unknown consequences. Subjects repeatedly receive feedback from past choices about the consequences of unfamiliar alternatives but this feedback is aggregated over states. Varying the payoffs attached to the various alternatives in various states allows us to test whether unfamiliar alternatives are discounted and whether subjects' use of feedback is better explained by similarity-based reinforcement learning models (in the spirit of the valuation equilibrium, Jehiel and Samet, 2007) or by some variant of Bayesian learning model. Our experimental data suggest that there is no discount attached to the unfamiliar alternatives and that similarity-based reinforcement learning models have a better explanatory power than their Bayesian counterparts. [ABSTRACT FROM AUTHOR]

We propose a new solution concept, called Context-Dependent Equilibrium Under Ambiguity (CD-EUA), for strategic games where players' beliefs may be influenced by exogenous context-related information. Players' beliefs about the strategic behavior of their opponents are represented by belief functions. The notion of belief functions allows us to combine exogenous context information in the spirit of Schelling (1960) with endogenous equilibrium beliefs about the opponents' behavior in analogy to the standard Nash equilibrium. For any finite strategic game, we prove existence of a CD-EUA for any context information and any degree of confidence in it. Moreover, we show continuity of the equilibrium correspondence. Finally, we illustrate how CD-EUA can be applied to different types of context information in games by explaining some stylized facts from experimental research on coordination. [ABSTRACT FROM AUTHOR]