THESIS
1999
iv, 128 leaves : ill. ; 30 cm
Abstract
Convergence of decisions is often observed in situations where individuals receive independent private signals and have access to the decisions made by preceding others, and the decisions are made in an exogenously ordered sequence. Bikhchandani, Hirshleifer and Welch (1992) proposed a herding model with an independent payoff structure for individuals (no externality) that explains the convergence of decisions by information cascade. Based on this model, the current thesis examines how well information signaling accounts for individuals' decisions in a group setting (Experiment 1) and in an equivalent logical task (Experiment 2). By enriching the information structure, the two experiments generalize earlier experimental studies of herd behavior and identify two factors responsible for t...[
Read more ]
Convergence of decisions is often observed in situations where individuals receive independent private signals and have access to the decisions made by preceding others, and the decisions are made in an exogenously ordered sequence. Bikhchandani, Hirshleifer and Welch (1992) proposed a herding model with an independent payoff structure for individuals (no externality) that explains the convergence of decisions by information cascade. Based on this model, the current thesis examines how well information signaling accounts for individuals' decisions in a group setting (Experiment 1) and in an equivalent logical task (Experiment 2). By enriching the information structure, the two experiments generalize earlier experimental studies of herd behavior and identify two factors responsible for the departure from the normative model. One concerns the violation of the assumption of common knowledge of rationality. The other involves a nonprobabilistic factor in the context, plausibly caused by the illusionary impact of the cascade. Alternative behavioral heuristics were found to outperform the normative model in describing the data. Implications to judgment and decision making and consumer learning are discussed.
Post a Comment