LAUSR

Decentralized sanctioning arises from a demand for governance that is not adequately provided by the state or another strong and centralized institution. While the dynamics of collective action and sanctioning have been well-examined theoretically, experimentally, and empirically, this work typically assumes community membership is a given. In selective or elite communities, pro-social behavior of one kind or another may be a prerequisite of community membership, which may create perverse incentives for the implementation of peer-sanctions. This article quantitatively examines this phenomenon in the case of professional ice hockey, a highly selective community where fist-fighting between players has long existed as a form of self-help for players to address rule infractions or violent play otherwise unaddressed by officials. An empirical examination of over 70 years of player statistics and play-by-play data from the National Hockey League shows not only the evolution of this system from one of peer-sanctioning to one of specialized-sanctioning, as might be predicted from experimental results showing the favorability and efficacy of more centralized punishment regimes, but also reveals how specialization has led to self-serving sanctions. Less-skilled players who are presumably hired to fight are disproportionately likely to participate in fights that appear to occur for non-retaliatory reasons, and more likely to fight one another in a bid to maintain their status and reputation as sanctioners, and consequently their membership in an elite community.