“In many areas of social life different parties interact under conditions of rivalry, striving for something that not all can obtain. Examples of such rivalries in the economic and political realms are R&D competition, promotion tournaments in internal labor markets, lobbying for government favors and electoral competition between political parties. As a result of such rivalries considerable resources are spent on activities that have no direct productive value. For example, […] previous to the adoption of auctions by the FCC, the real resources spent on filing applications for cell phone license lotteries (with an estimated market value of one billion dollars at that time) was about 400 million dollars. Extreme instances of rivalry are military conflicts and socio-political conflicts, like those that arise between parts of a country, when one of them is fighting for a different political status or independence, and those between ethnic groups. Actual conflicts of this type are often very costly, both in human lives and in material losses. […]
“In [many] rent-seeking experiments […] it is individuals who compete for a prize. In many naturally occurring situations, however, players are groups, since political parties, social movements, and associations like trade unions, lobbyists, terrorist groups etc. are invariably composed of more than one individual. Rent-seeking competition between groups rather than single players introduces an additional layer of complexity to the strategic characteristics of the interaction. Although groups clearly have the potential to be more powerful competitors than individual agents, they face internal coordination problems that may severely undermine their efficacy.
“[…] thus far it is poorly understood how human decision makers actually behave in simple collective rent-seeking contests. Consider a setting where all group members reap the benefits of success, while the likelihood of success depends on the efforts of individual group members. If formal enforcement measures are absent, the conflict parties effectively compete on the basis of voluntary contributions although informal sanctions against defectors, like social ostracism or mobbing, may help to overcome the inherent free-riding incentives. To date we have no systematic empirical evidence on how inter-group conflict is likely to evolve in such a setting.
“In the work we present here we use laboratory methods to study how conflict in contest games is influenced by parties being groups instead of individuals and by the existence of the possibility of punishment between members of a party. […] One can see this as a representation of a situation where the prize has a public good flavor for the successful party as is the case in some political confrontations in which all members of the winning party benefit from the outcome.
“Our results for the case without punishment show that expenditure levels in contests between groups are much higher than in contests between individuals, and both exceed equilibrium levels. On average, we observe that teams spend on conflict more than four times as much as predicted and about twice as much as single players. We also find that individual parties fighting against group parties invest similar levels to individual parties fighting against other individual parties. Group parties fighting against individual parties invest like group parties fighting against other groups.
“In contests with punishment opportunities expenditure levels are in turn much higher than in any of the treatments without punishment. In the final rounds of the experiment, investments in conflict are more than twice as high with punishment as without. The consequence is a large waste of resources: more than three quarters of the prize parties are fighting over are dissipated by direct conflict expenditures. However, to determine the true efficiency loss the costs imposed by punishment and the costs borne to punish others need to be added. These costs included, material losses are now 869 percent of the equilibrium level and rent dissipation is in excess of 100 percent. These results strongly contrast with those from those public goods experiments where punishment tends to enhance efficiency.”