This paper examines how a change in the generosity of one social assistance program generates spillovers onto other social assistance programs. We exploit an age discontinuity in the stringency of the 1993 Dutch disability reforms to estimate the causal effect of exit from disability insurance (DI) on participation in other social assistance programs. We find strong evidence of social support shopping: 43 percent of those induced to leave DI due to the reform receive an alternative form of social assistance two years after the implementation of the reform. As a result, for each Euro saved in DI benefits, the government has to spend an extra 60 cents in other social assistance programs. This crowd-out rate grows from 60% to 69% if we also take into account the response of the partners' of those affected by the DI reform. The crowd-out effect declines over time, but is still 25% eight years after the reform.
crowd-out spillover effects social insurance income assistance welfare regression discontinuity administrative data