Three Essays on Emotion, Coordination, and Economic Behavior

Abstract

Economic experiments provide controlled environments to examine the causality between economic variables, test theoretical frameworks, and offer insights for the development of new economic theory. In this dissertation, I employ laboratory experiments to study how behavioral factors influence individual decision-making and strategic interactions. This dissertation consists of three chapters. In the first chapter, "Minimum-effort game under stochastic monitoring", I design and experimentally test a coordination device "stochastic monitoring" in classical minimum-effort games, where group production is determined by the minimum effort provided. In these games, strategic uncertainty regarding others' contributions often results in inefficient, low-effort outcomes. Under stochastic monitoring, an automated supervisor monitors individual effort with a probability p and imposes a non-deterrent marginal fine c for deviations from the maximum effort level. I test the effectiveness of three treatments, Full-p, High-p, and Low-p, in improving coordination compared to a baseline when stochastic monitoring is absent. All treatments have the same expected marginal fine pc, but the values of p and c vary. Subjects were randomized into groups of four and interacted repeatedly. Stochastic monitoring was introduced after groups experienced coordination failure. The results show that stochastic monitoring significantly improves coordination across all treatments. Notably, the Low-p treatment (low probability of monitoring, high marginal fine) is the most effective, outperforming the Full-p treatment. Further analysis reveals that these improvements are not driven by direct feedback from punishment or individual traits. This research demonstrates that stochastic monitoring is an effective coordination device in environments where the production function has a "weakest-link" feature. It also shows that low-cost monitoring protocols (represented by low probability of monitoring) can achieve outcomes equal to, or better than, high-cost alternatives. In the second chapter, "Emotional echoes: Reference-dependent preferences and emotion dynamics" (joint with Clément Staner), we use a laboratory experiment to explore individual investment decisions when the decision-maker is frustrated. Traditional economic models often struggle to explain why unrelated past frustrating events influence subsequent decisions, even if wealth and beliefs remain unchanged. This chapter addresses the gap by investigating the dynamic role of emotions, specifically "frustration", in decision-making under uncertainty. We propose a theoretical framework where falling below a subjective reference point triggers frustration. Frustration accumulates over time and affects utility negatively through an emotional cost function. The model distinguishes between two environments: return-control (R), where investment increases the size of the reward, and probability-control (P), where investment increases the likelihood of success. In a within-subject experiment, subjects made investment decisions both before and after they experienced a frustrating event. Frustration was triggered via losing a lottery unrelated to the investments. We hypothesize that whether subjects increase or decrease investment depends on their sensitivity to frustration (curvature of the emotional cost function). Furthermore, ceteris paribus, subjects have a stronger incentive to invest in the P environments to increase the probability of removing the emotional cost. The results provide evidence for heterogeneous behavioral responses: Subjects with diminishing sensitivity to frustration escalate investment following a frustrating event, while those with increasing sensitivity de-escalate. These individual-level investment changes offset each other in aggregate data, which can potentially explain the lack of consensus in the existing "sunk cost" literature. While we confirm the directional predictions regarding investment changes across environments, the differences in magnitude of these changes between R and P are not statistically significant. The last chapter, "Emotions and market activity: Cause or consequence?" (joint with Daniel Gotsman and Charles N. Noussair), disentangles a long-standing debate about the causality between emotions and stock market movements. Despite widespread anecdote associating "exuberance" with high prices, or "fear" with low prices, the direction of causality remains poorly understood. We use a laboratory experiment to exogenously induce emotions of Happy, Fearful, or Neutral among traders. Subjects watched video through virtual reality (VR) headsets to induce the targeted emotional states; traders in the same market watched the same video. They then traded assets with stochastic dividends in a continuous double-auction market. This method allows us to directly compare the price and trading volume across markets under different emotional states, while holding other variables constant. The results show that exogenously inducing emotions does not affect asset pricing or trading volume, both on the session level and on the individual level. However, we find a robust reverse relationship: Market outcomes significantly affect subsequent self-reported emotional states. Specifically, traders who earned more experience significant increase in Joviality and decrease in Fear. These results suggest that the market euphoria observed in the field when price increases is a consequence of traders accruing greater earnings, rather than a cause of increased price.