Algorithmic Fairness and Digital Financial Stress: Evidence from AI-Driven E-Commerce Platforms in OECD Economies

Authors:
Open Access
Publisher website
Google Scholar
Abstract
This study examines the role of algorithmic fairness in alleviating digital financial stress among consumers across OECD countries, utilizing panel data spanning from 2010 to 2023. By introducing a digital financial stress index—constructed from indicators such as household credit dependence, digital debt penetration, digital default rates, and financial complaint frequencies—the research quantitatively captures consumer financial anxieties within AI-driven e-commerce platforms. Employing two-way fixed-effects regression and system-GMM methods to address endogeneity and dynamic panel biases, findings robustly indicate that increased algorithmic fairness significantly reduces digital financial stress. Furthermore, the moderating analysis highlights digital literacy as a critical factor amplifying fairness effectiveness, revealing that digitally proficient societies derive greater psychological and economic benefits from equitable algorithmic practices. These results contribute to existing scholarship by extending discussions of algorithmic ethics from individual-level analyses to a macroeconomic perspective. Ultimately, this research underscores algorithmic fairness as a crucial policy lever for promoting consumer welfare, calling for integrated national strategies encompassing ethical algorithm governance alongside enhanced digital education initiatives within OECD contexts.
Funding Information
  • The Youth Fund Project of Humanities and Social Sciences of the Ministry of Education (24YJC790194, 2024GGJS159, 2023BJJ010, 252400411114)
  • Foundation for the University Youth Key Teacher Training Plan by the Ministry of Henan (24YJC790194, 2024GGJS159, 2023BJJ010, 252400411114)
  • Philosophy and Social Science Planing Project of Henan (24YJC790194, 2024GGJS159, 2023BJJ010, 252400411114)
  • Henan Province Soft Science Research Program Project (24YJC790194, 2024GGJS159, 2023BJJ010, 252400411114)

This publication has 91 references indexed in Scilit: