Causal effects are identified based on a reform of the Norwegian early retirement program, which entailed that adjacent birth cohorts were exposed to completely different work incentives from age 62. The reform removed a strict retirement earnings test such that pension wealth was redistributed from early to late retirees. Given the pre-existing employment and earnings patterns, this implied a considerable rise in old-age income inequality. In principle, this could have been offset by changes in the labor supply. We estimate that the reform triggered a 42% increase in hours worked during the ages covered by early retirement options. However, as the labor supply responses were of similar magnitudes across the earnings distribution, they did little to offset the rise in inequality. As measured by the Gini coefficient, inequality in overall old-age income rose by approximately 0.03 (17%).