A Theory of Noncontributory Pension Design

Noncontributory subsidies for the old poor (first-pillar pensions) affect the welfare of hundreds of millions around the world. Their benevolent rationale is to redistribute progressively, subject to efficiency considerations. This paper focuses on a critical efficiency issue: first pillars may affect another, even bigger program, namely contributory pensions for the middle classes, by inducing a reduction in the density of contributions. A major source of concern with contributory pensions in emerging economies is that the total replacement rate is too small for participants with low density, which are prevalent. The paper develops a model where density of contribution is endogenous, because for a substantial subset of jobs, the State is unable or unwilling to impose a mandate to contribute. Thus, the job selection decision is bundled with a saving decision. The first finding is that bundling modifies the effective rate of return on contributions, raising it without bound as earnings in uncovered jobs become smaller (relative to earnings in covered jobs). Another finding is that the standard designs of first-pillar pensions reduce the equilibrium density of contributions. Thus, standard first-pillar designs do crowd out contributory pensions for the middle classes. The paper then analyzes two second-generation designs. The "proportional" minimum pension is found to create horizontal inequity and inefficiency. In contrast, a subsidy with a small withdrawal rate applied to contributory pensions minimizes the loss of contribution density. Optimal income taxation theory suggests that the latter also provides the most efficient progressive redistribution.