In an era of escalating retirement crises, conventional methods for assessing savings adequacy remain plagued by exorbitant costs, invasive privacy demands, reliance on sensitive financial details, and a glaring disregard for personal lifestyle choices. This study presents a transformative, scalable, and privacy-preserving solution: a supervised machine learning framework that accurately predicts individualized retirement readiness among middle-aged Hong Kong workers (aged 35–55; N = 278, oversampled to 402). Leveraging validated ground-truth labels from an interactive Personalized Pension Projection tool and low-cost psychological and sociodemographic predictors—carefully selected through the established Capacity-Willingness-Opportunity framework—we evaluated six models. The Random Forest algorithm emerged as the standout performer, achieving 86.6% accuracy, 90.4% precision in identifying inadequate savings, and 91.0% recall for adequate savings—surpassing Light Gradient Boosting Machine, SVM, and other competitors. Ablation studies repeatedly underscored retirement goal clarity as the most influential predictor. These results equip policymakers with a precise, affordable, and personalized screening tool to detect retirement shortfalls early, while spotlighting retirement goal clarity as a critical focus for impactful future interventions.

Authors:
Alex Yue Feng Zhu, The Education University of Hong Kong, Hong Kong

About the Presenter(s)
Dr Alex Yue Feng Zhu
Assistant Professor, The EDUHK
Research interests: behavioral finance, personal finance, experimental finance
Projects: IFEC(HK) financial education research fund (2022-2023), HK Gov Public Policy Research Scheme (2023-2025)

See this presentation on the full schedule – Sunday Schedule

Bookmark