"Paying Too Much Attention: Theory and Evidence" New! with Hyeonggyun Ko and Adnan Mahmood
When the stakes rise, most people reasonably look at more information so they can decide confidently. But their confidence improves less than the evidence justifies. Because they treat each piece of evidence as weaker than it really is, they keep looking until they feel sure enough—and so pay more attention than the stakes warrant. In a lab experiment, that over-attention cost them about 25% of their potential earnings.
"Costly Information Acquisition and Information Representation" with Suha Kim and Hyeonggyun Ko (Draft coming soon!)
When people decide how much to learn, we may expect that they learn the same amount whether they are asked how accurate to be or how many samples to observe. But asking them differently may change how much they learn. If so, we can infer the wrong learning behavior depending on how we ask them. In a lab experiment, a majority have a consistent learning behavior, and most prefer to learn by choosing how many samples to observe. Even then, changing the question can reduce how much they learn.
"Yquilibrium" with Jacob K. Goeree and Brett Williams (New draft coming soon!)
General equilibrium relies on convexity: without it, Walrasian equilibria typically do not exist. We propose an alternative solution concept, Yquilibrium, that applies to convex and non-convex economies alike. Yquilibria minimize the Economy’s Potential, and require utility clearing rather than market clearing. Laboratory markets confirm Yquilibrium predictions
"Overinference from Weak and Correlated Signals"
Decision-makers exhibit overweighting bias and correlation neglect, leading to extreme beliefs. Directly improving for overweighting bias affects correlation neglect and vice versa.
"Overconfidence and Social Learning" with Adnan Mahmood
There are distinct effects of confidence on welfare that depends on whether one exhibits relative or absolute confidence about their signal accuracy. Relative overconfidence improves welfare while absolute overconfidence worsens welfare after realizing ones true accuracy.
"Sharing information: Could Experts Consolidate?" with Adnan Mahmood
No, with MLRP; yes, with strong complementary information.
"An Efficient Value on Games with Graph"
An efficient value on cooperative games with graphs is proposed and axiomatically characterized. An example on disaster fund allocation is given.
"Anchored Updating from Recommendations" with Hyoeun Park (Journal of Mathematical Economics)
Individuals update their beliefs from recommendations on a menu instead of noisy signals. A belief updating rule that satisfies order independence of recommendations is proposed and characterized in this environment.
"Designing Markets" with Jacob K. Goeree and Brett Williams
"Y Dynamics" with Jacob K. Goeree and Brett Williams