本文主要針對確定提撥制退休金與確定給付制退休金計畫進行探討,以勞工退休金條例做為實例,討論個人帳戶的資產配置情形,其中可以包含N種投資標的,針對傳統資產配置理論之不足,提出以粒子群最佳化執行資產配置,並與基因演算配置做比較;最後,以精算模型探討退休金的適足性,所有的精算假設均為隨機,包括模擬薪資成長率、通貨膨脹率、預定利率、投資標的投資報酬率等情況,就資產配置方法的退休金的適足性,以個人帳戶累積金額以及所得替代率加以評估,針對個人的風險忍受度架構一個適用於每個人的資產配置模型。
Given the current benefits structure of the Taiwan Employee Retirement Income Security Act (TERISA), two pension plans (defined contribution and defined benefit) are examined to discuss the asset allocation of personal accounts, which may include N target investment strategies. The study proposes Particle Swarm Optimization to be compared with Genetic Algorithms for carrying out asset allocation and the findings indicate that Particle Swarm Optimization outperforms Genetic Algorithms in both solution quality and calculating time. The adequacy of the pension plans is examined by an actuarial model based on the hypothesis that all the variables are set as random, including simulated salary growth rate, inflation rate, interest rate, and investment return rate. The accumulated value and income-replacement ratio of personal accounts are further evaluated to construct an asset allocation model adequate for everyone according to each individual’s risk tolerance level.