Improving pension product design

Pension products characterized by linking an individual's savings directly to market returns represent the most popular, growing pension domain globally. These products are widely sold in contribution-defined pension schemes, labor market pensions, and individual schemes. However, available products are designed with a tendency to assume greater risk the longer it is until retirement, but are not adjusted to individual preferences and circumstances. This paper develops an optimal asset allocation strategy for a defined contribution plan by adjusting to individual needs, such that the expected utility of retirement benefits is maximized. An asset allocation strategy should not only depend on the plan member's age (or time left to retirement), nor only on her risk preferences, but should capture personal characteristics. Among other factors, we include current wealth, expected lifetime salary progression, expected social benefits, choice of assets, type of retirement distribution schedules, bequest motive and life insurance. The problem is solved via a model that combines two optimization technologies: stochastic control and multi-stage stochastic linear programming (SLP). As an example of an optimal pension product design, we present the operations research methods, which have potential to stimulate new thinking and add to actuarial practice.