September 14, 2018, 11:30–12:30
Room MF 323
IAST General Seminar
Biases in decision making are ubiquitous and have been documented through extensive experimental research. Field studies have replicated many of those biases, and sometimes find that bias decreases with relevant task-specific experience. It remains an open question, however, whether such debiasing occurs because of learning as opposed to selection at the population level. Answering this question requires time-varying data on individual decision making over a reasonably long period of time, which is difficult and costly to generate both in the lab and field. To address those limitations, we leverage the full record of play at the dominant online poker website in India -- a stable field environment where players have to make decisions under uncertainty that carry meaningful monetary consequences. We identify bias in decision making based on how players shift behavior as a response to exogenous, marginal wins and losses. We establish that, on average, this bias is substantial in the population of players. We propose an empirical strategy aimed at testing whether debiasing actually occurs as players get more experienced, and assess the respective contributions of selection and learning in this process.