Economic Consequences of Manipulation of Social Insurance Benefits

Abstract

This paper examines the economic consequences of manipulation of social insurance benefits. Using administrative data of the public long-term care insurance (LTCI) in Japan, we document novel discontinuity and bunching in the distribution of health scores that determine benefit levels for LTCI. The observed distribution suggests that LTCI recipients tend to receive more generous benefits than they should because medical examiners manipulate recipients' health score. To quantify the impact of manipulation on long-term care (LTC) expenditures, we develop partial identification and nonparametric estimation methods that allow for flexible restrictions on counterfactual distribution. Our baseline estimation indicates that the manipulation increases monthly LTC expenditures per recipient by 60.2-227.9 USD (3.7-15.5%). We also find that the lower bound on manipulation effects is robust to various restrictions.