There is evidence that people do not fully take into account how other people’s actions are contingent on these others’ information. This paper defines and applies a new equilibrium concept in games with private information, "cursed equilibrium", which assumes that each player correctly predicts the distribution of other players’ actions, but underestimates the degree to which these actions are correlated with these other players’ information. We apply the concept to common-values auctions, where cursed equilibrium captures the widely observed phenomenon of the winner’s curse. We also show how cursed equilibrium predicts other empirically observed phenomena, such as trade in adverse-selection settings where conventional analysis predicts no trade, and "naïve" voting in elections and juries where rational-choice models predict that voters fully take into account the informational content in being pivotal.

This dissertation uses controlled experiments to understand why people fall shortof making rational decisions in simple financial decision-making situations, how we can restore rationality through simple interventions, and how providing people with substantial opportunities to learn about their environment might lead them away from making rational decisions.

In the first chapter, joint with Guangli Zhang, we study the sources of suboptimalallocations observed in credit card repayments using a diagnostic laboratory experiment. We find that optimization ability and limited attention are jointly insufficient to explain the puzzle. Moving beyond existing results, we find that the inherent negative frame of the debt payment problem interferes with subjects’ ability to optimize and hinders learning. We show that subjects predominantly rely on the irrelevant balance information while forming their decisions, regardless of how vividly the balance information is displayed. Using additional treatments, we find that the debt frame increases subjects’ focus on the irrelevant balance information.

In the second chapter, joint with Guangli Zhang, we study what type of interventionswould be effective in eliminating simple arbitrage failures in repayments. We construct a simple repayment environment in the laboratory and test the role of a set of behavioral mechanisms that would directly inform the design of consumer protection policies. We find that providing salient interest rate disclosure has no effect while disclosing the interest rate in a fee format has modest effects. On the other hand, providing an opportunity to purchase automated financial advice reveals that subjects are predominantly aware of their choice inefficiencies and are relatively good at gauging the extent of their mistakes and using financial advice. Our results suggest that promoting and subsidizing consumer financial technology applications that provide automated financial advice would be a substantially more effective way of protecting consumers from simple arbitrage failures than conventional disclosure policies.

In the final chapter, I study how people learn about their environment when theirsubjective understanding of the environment, their mental model, is misspecified. I use people’s tendency to hold optimistic beliefs about their abilities to generate a significant amount of model misspecification and investigate the implications of overconfidence as a misspecified mental model on learning about own ability and a fundamental. Consistent with the theoretical predictions, overconfident subjects develop pessimistic beliefs about the fundamental and take growingly suboptimal actions. Inconsistent with the theoretical predictions, I find endogenous feedback does not exacerbate the extent of suboptimal behavior. Investigating how subjects learn about their own ability reveals that abundant feedback ”weakens” misspecified mental models. The ”weakening” of mental models is more pronounced with endogenous feedback and explains why endogenous feedback may not exacerbate the extent of suboptimal behavior.

This dissertation consists of two essays studying decision making in FinTech, and one essay on decision making in education.

The first chapter explores swapping decisions on decentralized exchanges (DEX). With the increasing adoption of DEX platforms, new Layer 2 (L2) blockchain alternatives offer better scalability and lower fees than the Ethereum blockchain (L1), yet the relative security of L2 is uncertain. Using a structural model and a novel dataset, we estimate investor preferences for blockchain security between mainnet (L1), and two main L2 networks, Polygon and Optimism. We find that traders anticipate a 0.68% (3.29%) chance of losing transaction value when trading on Polygon (Optimism) compared to L1, significantly higher than the transaction fees (0.01%-0.3%) charged on each trade. Our findings provide empirical evidence of the trade-off between scalability, security, and decentralization, a major challenge for blockchain networks.

The second chapter investigates default settings on DEX. DEX prices update continuously after each swap, causing potential price shifts for users awaiting execution. Users set a slippage tolerance to limit acceptable price increases, but this can either expose them to sandwich attacks or cause transaction failures. We analyze the impact of slippage tolerance settings on the health of Uniswap and Sushiswap ecosystems. A recent Uniswap change replaced the static default slippage setting (0.5%) with a dynamic one based on market conditions to reduce sandwich attacks. We find that this change significantly reduces trader losses by approximately 54.7%, with an even more pronounced effect (90%) for traders following the default settings. We also propose further improvements for these settings.

The final chapter examines student decision-making, specifically the impact of a gamified leaderboard on engagement and procrastination. Procrastination is common among students, particularly with assignments. Gamification, incorporating game-like elements into education, shows promise in addressing this issue. Our results indicate that students in the treatment group complete assignments faster, suggesting the leaderboard positively influences study behaviors. While the overall class performance effect is not significant, transfer students and male students exposed to the leaderboard achieve higher course grades than their peers in the control group.