I work on microeconomic theory and behavioural economics, with a special focus on information economics.
I concluded my Ph.D. in economics at Columbia University in 2021.
A new equilibrium solution concept where players sequentially sample to resolve strategic uncertainty over their opponents' distribution of actions delivers predictions on the joint distribution of actions, beliefs, and decision times.
Choice mistakes are rare and off-path moves strongly predict subsequent deviations. In contrast, while beliefs systematically overestimate others' mistake rates, they fail to account for this correlation and severely underinfer from off-path choices.
A longitudinal randomised controlled trial on a movie recommendation platform identifies an informational channel of recommendations on consumption by examining how recommendations affect consumers' beliefs and how these determine consumption choices.
(An earlier version circulated with the title "The Economics of Recommender Systems: Evidence from a Field Experiment on MovieLens")
Even in the simplest sequential-sampling model, errors can be fast or slow: the error rate is generically monotone in decision time, and optimal stopping in consumer choice can rationalise both patterns and we characterise when either holds.
In a consumer-choice application, whether mistakes occur early or late depends on a product's popularity and price.
We leverage the fact that minor changes to relative incentives exert a disproportionately larger impact on choices in complex problems than in simpler ones to investigate how different features of the decision-making environment contribute to the complexity of choice across various domains of interest.
Differences between laboratory and online participants partly reflect differences in the opportunity cost of time: experimentally changing time costs shifts both decision speed and accuracy. These effects are concentrated at intermediate levels of task complexity, where additional deliberation is most productive.
The incentive level in games affects choices and beliefs about the opponent by inducing greater cognitive effort.
(An earlier version circulated with the title "The Effect of Incentives on Choices and Beliefs in Games: An Experiment")
Retractions — provision of indirect information — are treated as more complex than equivalent direct information and subjects infer less from them.
(An earlier version circulated as a NBER working paper with the title "Learning versus Unlearning: An Experiment on Retractions")
We provide a complete characterisation of the set of QRE in a class of binary-action games with a continuum of types, including global games, which we leverage to develop nonparametric tests.
The ability of platforms to bias their recommendations can lead to vertical foreclosure, but separating recommendation and production is not always welfare improving.
