Working Paper: CEPR ID: DP5551
Authors: Helmuth Cremer; Jean-Marie Lozachmeur; Pierre Pestieau
Abstract: In many countries pension systems involve some form of earnings test; i.e., an individual?s benefits are reduced if he has labour income. This paper examines whether or not such earning tests emerge when pension system and income tax are optimally designed. We use a simple model with individuals differing both in productivity and their health status. The working life of an individual has two 'endings': an official retirement age at which he starts drawing pension benefits (while possibly supplementing them with some labour income) and an effective age of retirement at which professional activity is completely given up. Weekly work time is endogenous, but constant in the period before official retirement and again constant (but possibly at a different level), after official retirement. Earnings tests mean that earnings are subject to a higher tax after official retirement than before. We show under which conditions earnings tests emerge both under a linear and under a nonlinear tax scheme. In particular, we show that earning tests will occur if heterogeneities in health or productivity are more significant after official retirement than before.
Keywords: earnings test; social security
JEL Codes: H21; H55; J26
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| greater dispersion in health status or productivity after retirement (J26) | earnings tests in pension systems (H55) |
| higher marginal tax on post-retirement earnings (J26) | earnings tests in pension systems (H55) |
| earnings tests in pension systems (H55) | discourage labor supply (J79) |
| increase in inequality among retirees (J26) | existence of earnings tests (H55) |
| variance in health or productivity increases post-retirement (J26) | higher marginal tax on earnings after retirement (J26) |