Manipulation, Selection and the Design of Targeted Social Insurance

This paper provides a sufficient statistics framework to study the design of optimal targeted social insurance in the presence of manipulation opportunities, through which individuals select into policies not intended for them. We apply our framework to Italian unemployment insurance (UI), which features a discontinuous coverage increase around an age-at-layoff threshold. Using novel bunching techniques, we document pervasive manipulation and a substantial increase in benefit receipts. However, most of this increase is mechanically due to higher coverage, while the implied moral hazard response is modest. We connect these findings to our theory and discuss how they affect welfare conclusions