How much do we learn? Measuring symmetric and asymmetric deviations from Bayesian updating through choices

B-Tier
Journal: Quantitative Economics
Year: 2025
Volume: 16
Issue: 1
Pages: 329-365

Authors (4)

Ilke Aydogan (not in RePEc) Aurélien Baillon (EMLYON Business School) Emmanuel Kemel (not in RePEc) Chen Li (not in RePEc)

Score contribution per author:

0.503 = (α=2.01 / 4 authors) × 1.0x B-tier

α: calibrated so average coauthorship-adjusted count equals average raw count

Abstract

Belief‐updating biases hinder the correction of inaccurate beliefs and lead to suboptimal decisions. We complement Rabin and Schrag's (1999) portable extension of the Bayesian model by including conservatism in addition to confirmatory bias. Additionally, we show how to identify these two forms of biases from choices. In an experiment, we found that the subjects exhibited confirmatory bias by misreading 19% of the signals that contradicted their priors. They were also conservative and acted as if they missed 28% of the signals.

Technical Details

RePEc Handle
repec:wly:quante:v:16:y:2025:i:1:p:329-365
Journal Field
General
Author Count
4
Added to Database
2026-01-24