"Do digital reputation systems correct for market failures in information asymmetry, or do they create new inefficiencies through algorithmic discrimination?"

Abstract

Digital reputation systems, such as seller feedback scores and user ratings, have become ubiquitous mechanisms for building trust in online markets. While scholarly discourse often frames these systems as remedies for the “lemons problem”, a classic market failure rooted in information asymmetry. This dissertation investigates whether digital reputation systems (e.g., seller feedback scores) correct market failures in information asymmetry or create new inefficiencies through algorithmic discrimination. Framing information asymmetry as a classic market failure (Akerlof's "lemons problem") in markets like e-commerce and freelance labor, it draws on Belle Flamme and Peitz to argue these systems can be effective, first-best solutions. However, the central argument reveals a paradox: by reviewing empirical work by Liu and Sefers and theoretical models by Rhodes and Zhou, the study demonstrates how algorithmic design inadvertently generates new inefficiencies—namely, statistical discrimination and herding. The dissertation concludes that these outcomes are co-produced by the same logic: reputation systems are powerful tools but not a panacea, demanding platform governance that balances trust-building with social equity.