Leveling up? An inter-neighborhood experiment on parochialism and the efficiency of multi-level public goods provision

Johannes Lohse, Timo Goeschl, Martin Kesternich, Daniel Roemer, Carlo Gallier

Research output: Working paper/PreprintWorking paper


Many public goods can be provided at different spatial levels. Evidence from social identity theory and in-group favoritism raises the possibility that where higher-level provision is more efficient, subjects’ narrow concern for local outcomes (parochialism) could harm efficiency. Building on the experimental paradigm of multi-level public good games and the ‘neighborhood attachment’ concept, we conduct an artefactual field experiment with 600 participants in a setting conducive to parochial behavior. In an inter-neighborhood intraregion design, subjects allocate an endowment between a personal account, a local, and a regional public good account. The between-subjects design varies across two dimensions: One informs subjects that the smaller local group consists of members from their own neighborhood (‘neighbors’). The other varies the relative productivity at the two public goods provision levels. We find evidence for parochialism, but contrary to our hypothesis, parochialism does not interfere with efficiency: The average subject responds to a change in relative productivities at the local and regional level in the same way, whether aware of their neighbors’ presence in the small group or not. The results even hold for subjects with above-median neighborhood attachment and subjects primed on neighborhood attachment.
Original languageEnglish
PublisherCentre for European Economic Research (ZEW)
Number of pages46
Publication statusPublished - 3 Mar 2017

Publication series

NameDiscussion Paper Series


  • Social Identity
  • Parochialism
  • Multi-Level Public Goods
  • Artefactual Field Experiment


Dive into the research topics of 'Leveling up? An inter-neighborhood experiment on parochialism and the efficiency of multi-level public goods provision'. Together they form a unique fingerprint.

Cite this