Couples' retirement under individual pension design: A regression discontinuity study for France

Retirement policies are individually designed, but the majority of older workers are partnered, and are likely to coordinate their employment decisions with their spouse. The goal of this study is to estimate the direct and indirect (via the spouse) effects of a pioneer French pension reform on both spouses' retirement decision. The extent of the reform varies by birth year, which enables us to identify its retirement effects on both spouses, since the husband is, on average, 2 years older than the wife. We use labor-force survey data to implement a sharp regression-discontinuity framework, in which the running variable is the distance of the individual birth month to a certain reference month, as well as an incremental differences-in-differences approach. We find a significant drop in each spouse's probability of retirement. The husband's retirement probability also drops immediately by 1 percentage point if the wife is affected by the reform, while her retirement probability does not respond immediately if he is affected.