The pool of early retirees is characterized by a large heterogeneity along several criteria. The present paper focuses on the key distinction between those in forced early retirement and those who retire early by individual choice. We start by estimating a retirement probit model for older workers in Belgium. Based on these estimates, we then perform micro-simulations relating to a hypothetical actuarial reform of a pension system, i.e., a reform imposing on average actuarial neutrality with respect to the time of retirement. We explore two scenarios, one where the entire population is subjected to the actuarial system, and one where a duly screened sub-sample of the unemployed is shielded against these actuarial adjustment factors, a group we call the truly unemployed. We evaluate the impact on the average retirement age, the pension budgets as well as indicators of redistribution within the group of the elderly. We find that the extra budgetary gain of exposing this subgroup to the full-blown reform is modest, while the distributional cost is rather high. Our results thus comfort the idea that the budgetary cost of a focused unemployment system are moderate, and that returning the unemployment insurance to its primary role might be a desirable strategy.