Hong Kong’s Mandatory Provident Fund (MPF) system plays a critical role in securing retirement outcomes for an aging population, requiring portfolios that deliver stable, long-term growth under regulatory and cost constraints. However, MPF schemes face structural challenges in managing downside risk: dynamic rebalancing and derivatives are costly, operationally complex, and constrained by Mandatory Provident Fund Schemes Authority (MPFA) regulations, while the predominantly passive investment philosophy limits active risk-hedging tools. As a result, protecting MPF portfolios against market downturns without increasing costs or regulatory burden remains difficult.
To address these challenges, the project proposes a passive, low-cost, cross-sectional downside-risk management framework that redesigns MPF equity building blocks rather than hedging market risk directly. The approach combines two complementary signals: a learning-based method that uses option-implied volatility surfaces and neural networks to estimate forward-looking crash probabilities, and a theory-driven robustness method that derives conservative worst-case performance bounds consistent with option prices. Together, these signals form a transparent, governance-friendly scorecard to identify fragile versus resilient stocks. Building on this, the project will launch the HKAIFT SafeSelect Index, a rules-based passive index selecting the most resilient 80% of stocks, alongside benchmark-aware overlays with tracking-error and turnover controls suitable for MPF mandates.
Complementing portfolio innovations, the project will enhance an AI-driven MPF Chatbot to improve retirement decision-making and financial literacy among MPF members, particularly non-digital natives. The chatbot integrates multilingual, voice-enabled access with validated data retrieval, personalized education, fund comparison, and asset allocation guidance, reinforced by an “A.I. vs. A.I.” validation layer to ensure compliance, accuracy, and responsible GenAI. Following successful prototyping through a FinTech Social Hackathon, the next phase will operationalize the downside-risk signals, launch the HKAIFT SafeSelect Index, and expand the chatbot with stress-period analytics. Through collaboration with MPF trustees, operators, and regulators, the project aims to deliver scalable, regulatory-compliant tools that strengthen downside protection and support sustainable retirement outcomes.
