This paper investigates the consequences of relying on assets accumulated in a defined contribution pension plan compared to an annuity based on salary from a defined benefit plan. Although a defined contribution plan varies with asset returns, it may be more desirable than a defined benefit plan when wage variability and job turnover are adequately considered. It is found that both job separation rates and wage variance increased in the 1990s. The new calibrations of these variables are used in a life-cycle model where a worker chooses between a defined benefit and a defined contribution plan. It is shown that the increase in job turnover made defined contribution the dominant pension plan.

Many social security systems face high and escalating disability costs. In Chile's new system, the disability assessment procedure includes participation by private pension funds (AFPs) and insurance companies, who finance the benefit, have a direct pecuniary interest in controlling costs and are able to pursue this objective by helping to set criteria and providing countervailing information. We hypothesize that these procedures and incentives will keep costs low, by cutting the incidence of successful claims. Using the Cox proportional hazard model and a retrospective sample of new and old-system affiliates (EPS, 2002), we find that disability hazard rates are only 20–35% as high in the new system as in the old traditional system. Analysis of mortality rates suggests that the new system has accurately targeted individuals with more severe medical problems.

Conditional indexation has recently attracted interest with pension funds that are looking for a middle way between defined benefit and defined contribution. In this paper, we analyze conditional indexation schemes from a life-cycle investment perspective. Welfare analysis is applied to investigate the performance of such schemes relative to alternative investment strategies such as fixed-mix policies. We carry out this analysis in the context of a broad family of utility functions, which takes into account the possible presence of two benchmark levels corresponding to a minimum guaranty and to full indexation, respectively. For the purpose of comparability, we construct a self-financing continuous-time implementation of the conditional indexation scheme. The implementation involves continual adjustment of the parameters of the contingent claim representing final payoff. Our findings indicate that, in situations where large weight is placed on the benchmark levels, conditional indexation is fairly close to being optimal.

Evidence is presented for a model in which wage growth is positively correlated with equity returns after a time lag of 1–3 years. This model is used to derive the risk premium on an asset which provides a cash flow linked to final salary. Using historic UK data, it is estimated that this risk premium is 0.5% per annum, a much smaller figure than that normally assumed for the equity market. This result has implications for the discount rate that should be used to derive the fair value of final salary pension liabilities.

As life tables continue to lengthen, a world-wide paradigm shift to defined contribution – but undefined rewards – has left pensioners exposed to performance, credit, and longevity risk. However, relatively risk-free defined benefit schemes can be designed off the back of high-grade debt issuance programmes by public long-term asset vehicles, as an infrastructure–retirement double play. Derivatives can be used to enhance coupons and to correctly align risk preferences as between income while still alive and bequests. Variable lifetime reinvested coupon options and annuity swaps utilize market pricing to provide unambiguous pricing benchmarks and a necessary underpinning of lifecycle planning certainty. The result is a flexible mix of private and public provision of old age income assurance, one that exploits the externalities of a well-designed system of public debt.