Yen Ling Tan

This paper studies how cursedness—the tendency to neglect how other people's strategies depend on their private information—affects trade in a takeover game with one buyer and seller. I apply the Cursed Sequential Equilibrium concept, showing that information transmission and allocative efficiency depend on the degree of information asymmetry and size of the stakes. Finally, this paper discusses two experiments to test the model predictions and explore whether experience in different roles impacts cursed behavior.
Presentations: 2nd European Economic Review Summer School in Experimental and Behavioral Economics

We study how cursedness, the tendency to neglect how other people's strategies depend on their private information, affects information transmission in Spence's job market signaling game. We characterize the Cursed Sequential Equilibrium and show that as players become more cursed, the worker obtains less education—a costly signal that does not enhance productivity—suggesting that cursedness improves the efficiency of information transmission. However, this efficiency improvement depends on the richness of the message space. Revisiting the job market signaling experiment by Kübler, Müller, and Normann (2008), we find supportive evidence for our theory.
Presentations: University of Virginia Theory/Experimental Workshop

Inspired by the ride-sharing market in New Zealand—with Uber and Zoomy offering respectively a fixed price and an estimated price range per ride—we ask ourselves if competing platforms could deliberately offer distinct pricing schemes aimed at matching riders and drivers with different levels of ambiguity tolerance to gain market shares. We calibrate the distribution of ambiguity attitudes using data from a suitable preliminary laboratory experiment. Our duopoly pricing model predicts that, in spite of the realistic asymmetric distribution of ambiguity-loving versus ambiguity-averse riders and drivers, both platforms can coexist in the market in equilibrium. Ambiguity-loving users are attracted to the price range offers, whereas ambiguity-averse users shy away from them. In equilibrium, drivers from both platforms extract rents, so do riders who accept price range offers. However, all rents are successfully extracted away from riders who accept fixed price offers.
Presentations: Economic Science Association (ESA) 2024 North American Meeting, Allied Social Science Associations (ASSA) 2021 Annual Meeting, Australasian Economic Theory Workshop (AETW) 2020, Asia-Pacific Industrial Organization Conference (APIOC) 2019, Keio University Microeconomics Workshop, University of Nottingham, University of Auckland Center for Mathematical Social Science Lightning Talks, New Zealand Microeconomics Study Group Meeting

A binary lottery converts lab earnings into lottery tickets for a single cash prize. This conversion is predicted to induce risk neutrality because expected utility expressions are linear in probabilities, and binary lotteries with only two possible money outcomes provide no role for utility curvature. Binary lotteries are increasingly used in procedures that elicit subjective beliefs, which is convenient since standard methods are known to bias probability responses toward one half. Nevertheless, this trend is perplexing, since binary lotteries failed to induce risk neutrality in risk aversion assessment tasks (Kirchkamp et al., 2021). This paper investigates the performance of binary lotteries in simple attacker-defender games that are known to exhibit severe aversion to risk, i.e. when both the attacker and the defender have a completely safe strategy (Holt, Sahu, and Smith, 2022). The game used has the property that the unique Nash equilibrium stipulates equal probabilities for each decision for each player. The incidence of safe choices (to defer an attack or to go on high alert) is compared using binary lotteries in one treatment and unconverted cash payouts in another. The main result is that the binary lottery does not move the observed incidence of safe decisions toward the 50-50 risk neutrality predictions. We also consider a second pair of treatments with neutral terminology (row, column) instead of the hot terminology (defender, attacker, with a backstory about blocked or successful attacks). The conclusion is that the tendency to overuse safe strategies (defer attack or high alert) is not due to the hot terminology, and the binary lottery does not induce risk neutrality in either case. A second experiment will determine whether binary lotteries neutralize the well-known tendency for subjects to exhibit sharply increased levels of risk aversion in Holt-Laury choice tasks with scaled-up payoffs.

We study how different matching protocols used in laboratory experiments affect learning behavior in extensive games of imperfect information. In this experiment, we examine Selten’s Horse game under two matching protocols. In the fixed-role treatment, roles are assigned once and remain constant, while opponents are randomly matched in each repetition. In the switch-role treatment, both roles and opponents are randomly reassigned in every round. By leveraging different payoff structures, we disentangle the interactions between matching protocols and equilibrium selection.