This paper examines how raising the pension eligibility age affects retirement behavior in Japan. Using administrative tax data, we analyze a reform that increased the eligibility age for pension benefits from 60 to 61 for certain cohorts. Leveraging a difference-in-differences approach, we find that affected cohorts increased their employment rate at age 60 by approximately 5 percentage points relative to unaffected cohorts. Moreover, the employment effects persisted beyond the new eligibility threshold: affected cohorts remained more likely to work between ages 61 and 64, even after becoming eligible for pension benefits. As a result, the reform induced roughly 60 percent of individuals who otherwise would have exited the labor force to work approximately three additional years on average. This persistent effect is unlikely to be driven by income or wealth effects. Instead, it reflects two key features of Japan’s labor market institutions at older ages: (1) workers are mandated to re-sign new contracts at age 60 to continue employment, often involving substantial changes in work arrangements; (2) strong employment protection rules apply until age 65. Our results suggest that policies aimed at extending the working lives of older individuals should not only address demand-side constraints but also facilitate transitions to new job arrangements later in life.

This paper examines the effects of Japan’s old-age pension earnings test on the labor supply of elderly men, who were eligible for pension between ages 61 and 63, and were subject to the earnings test until age 65. Using administrative residential tax records from 25 municipalities, we analyze how individuals adjusted earnings in response to the kink points created by the earnings test and how work incentives changed following its removal. We document clear bunching at the first kink point, indicating strong behavioral responses to the implicit tax on earnings. Notably, this bunching persisted even after the earnings test was lifted at age 65, suggesting frictions in post-retirement earnings adjustment. The response is weaker among younger cohorts, who were subject to the test for shorter durations, and stronger among individuals with lower pension benefits, who faced less binding constraints. Overall, our results show that the earnings test substantially distorted labor supply at the intensive margin, but these distortions attenuate when thresholds become more generous.

Using a mandatory retirement policy in China, which requires female workers to leave their current job at age 50, and the provincial unemployment rate at the time of retirement, we examine the labor market outcomes of young retirees over the business cycle. We find that the costs of retiring in a weak labor market are substantial. Women who reach the statutory retirement age during periods of above-median unemployment are 8 percentage points (or about 13 percent) less likely to work between the ages of 50 and 58. Women with higher net family assets are relatively more sensitive to high unemployment at the statutory retirement age. Although consumption does not fall significantly, affected women’s households reduce spending on other items such as transfers.