(joint with Mohammed Abdellaoui and Brian Hill)
(joint with Mohammed Abdellaoui and Brian Hill)
Abstract: One popular approach to ambiguity is the smooth ambiguity model, which adopts a two-stage perspective and captures ambiguity attitudes by a utility transformation function. Another approach, also applicable in a two-stage framework, uses probability weighting to reflect ambiguity attitudes. Working with a specification that combines the features of both approaches, we elicit both the utility and probability weighting drivers of ambiguity attitudes. Our experiment reveals the relevance of both factors to describing ambiguity attitudes. We also report evidence of stakes-dependence of ambiguity attitudes, as implied by utility-driven approaches.
Abstract: One popular approach to ambiguity is the smooth ambiguity model, which adopts a two-stage perspective and captures ambiguity attitudes by a utility transformation function. Another approach, also applicable in a two-stage framework, uses probability weighting to reflect ambiguity attitudes. Working with a specification that combines the features of both approaches, we elicit both the utility and probability weighting drivers of ambiguity attitudes. Our experiment reveals the relevance of both factors to describing ambiguity attitudes. We also report evidence of stakes-dependence of ambiguity attitudes, as implied by utility-driven approaches.
Ignorance is Bliss? Domain Dependence of Temporal Risk Preferences
Ignorance is Bliss? Domain Dependence of Temporal Risk Preferences
(Job Market Paper; draft coming soon!)
(Job Market Paper; draft coming soon!)
Abstract: We report the results of an experimental investigation of preferences for noninstrumental information in the context of temporal risk resolution. Operating in the domain of both gains and losses, we fix the date of realization of the outcome and ask subjects to choose between two-stage lotteries that resolve completely in the present, gradually, or completely later. We find a predominant preference for early resolution and no preference for one-shot resolution in both domains, at the aggregate level. However, while the preference to know early becomes stronger at higher winning probabilities, the opposite is true for losses—higher the probability of losing, stronger the tendency to know later. At the individual level, we observe greater heterogeneity when facing losses, driven by a minority of subjects who prefer delayed resolution, an example of information avoidance. To refine our understanding of the heterogeneity in preferences, we carry out latent class analysis in each domain, adapting the framework of recursive expected utility.
Abstract: We report the results of an experimental investigation of preferences for noninstrumental information in the context of temporal risk resolution. Operating in the domain of both gains and losses, we fix the date of realization of the outcome and ask subjects to choose between two-stage lotteries that resolve completely in the present, gradually, or completely later. We find a predominant preference for early resolution and no preference for one-shot resolution in both domains, at the aggregate level. However, while the preference to know early becomes stronger at higher winning probabilities, the opposite is true for losses—higher the probability of losing, stronger the tendency to know later. At the individual level, we observe greater heterogeneity when facing losses, driven by a minority of subjects who prefer delayed resolution, an example of information avoidance. To refine our understanding of the heterogeneity in preferences, we carry out latent class analysis in each domain, adapting the framework of recursive expected utility.
Attitudes towards Information and Uncertainty
Attitudes towards Information and Uncertainty
(joint with Emmanuel Kemel)
(joint with Emmanuel Kemel)