Behaviourally adaptive optimization strategy for retirement wealth allocation under uncertainty

• Published in: Scientific African Vol. 29; p. e02873
• Main Authors: Assabil, Samuel Essamuah, Abubakar, Ali
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.09.2025; Elsevier
• Subjects: Life insurance; Personal wealth allocation; Retirement planning; Risk-free investment; Risky investment; Personal wealth allocation; Risk-free investment; Risky investment; Life insurance; Retirement planning
• ISSN: 2468-2276; 2468-2276
• DOI: 10.1016/j.sciaf.2025.e02873

•Dynamic model adapts to market shifts and personal financial experiences.•Panic selling and loss aversion modeled to reflect crash-driven volatility.•Shift from conservative to risky assets improves long-term wealth resilience. Optimal wealth allocation is essential for financial stability during retirement-a period marked by trade-offs among consumption, investment risk, life insurance, and bequest motives. Traditional static models often fail to capture the dynamic interplay among market volatility, income uncertainty, and behavioural tendencies such as loss aversion and panic selling. This study develops a dynamic, behaviourally adaptive optimisation framework for retirement planning under uncertainty. The problem is formulated as a dynamic stochastic multi-objective optimisation problem, wherein the investor seeks to maximise expected utility from consumption and bequest, subject to financial constraints and individual risk preferences. Wealth evolution is governed by recursive Bellman equations and solved using Monte Carlo dynamic programming, which captures non-linearities, volatility clustering, labour income shocks, and fat-tailed asset return distributions. The framework is validated through simulations of a hypothetical 40-year-old male investor with an initial wealth of $500,000 and an annual income of $100,000 over a 20-year horizon. The portfolio comprises stocks, bonds, real estate, gold, and commodities, each with empirically derived return-volatility profiles. Market uncertainty is modelled using a fat-tailed t-distribution with five degrees of freedom, and 50,000 Monte Carlo simulations are conducted to generate plausible market paths. Behavioural responses are modelled as a stochastic risk-aversion process influenced by recent financial experiences. Key downside risk measures, including the 5 % Value at Risk and Conditional Tail Expectation, are computed to assess portfolio resilience. The results indicate that the adaptive model outperforms static strategies by preserving wealth, absorbing employment shocks, and supporting long-term financial and bequest objectives.