Abstract
We test whether and, if so, how incentives to promote prosocial behavior affect the extent to which they spill over to subsequent charitable giving. To do so, we conduct a two-period framed field experiment to study repeated prosocial decisions of more than 700 participants. We vary how participants’ first prosocial behavior is incentivized by a wide range of interventions ranging from soft to hard paternalism. Our design allows us to decompose spillover effects into a pure spillover effect, which identifies the impact of previous prosocial behavior on subsequent donation decisions and crowding effects, which capture the extent to which the spillover effects are affected by the intervention exerted on the previous prosocial behavior. We find evidence for negative spillover effects. Participants donate less if they completed a prosocial task prior to the donation decision. Most importantly, we find that the spillover effects depend on how the initial prosocial behavior has been incentivized. Especially participants who are incentivized to donate through social comparisons are more willing to give to charity thereafter compared to participants whose initial prosocial behavior is incentivized by monetary rewards.