www.sciencemag.org/cgi/content/full/312/5770/108/DC1 Supporting Online Material for The Competitive Advantage of Sanctioning Institutions Özgür Gürerk, Bernd Irlenbusch, Bettina Rockenbach* *To whom correspondence should be addressed. E-mail: [email protected] Published 7 April 2006, Science 312, 108 (2006) DOI: 10.1126/science.1123633 This PDF file includes: Materials and Methods Figs. S1 and S2 Tables S1 ReferencesGürerk, Irlenbusch & Rockenbach 1 The Competitive Advantage of Sanctioning Institutions Supporting Online Material Özgür Gürerk*, Bernd Irlenbusch§ & Bettina Rockenbach* * University of Erfurt, Nordhäuser Straße 63, 99089 Erfurt, Germany, § London School of Economics, Houghton Street, London, WC2A 2AE, UK. Correspondence: Bettina Rockenbach, Chair in Microeconomics, University of Erfurt, Nordhäuser Straße 63, 99089 Erfurt, Germany. Phone: +49 361 7374521; fax: +49 361 7374529; e-mail: [email protected]ürerk, Irlenbusch & Rockenbach 2 SUPPORTING ONLINE MATERIAL Materials and Methods Methods 84 undergraduate students from the University of Erfurt voluntarily participated in the experiments. Special care was exerted to recruit students from many different disciplines to increase the likelihood that the subjects had never met before. Each participant was allowed to take part in one session only. In total 7 experimental sessions each involving 12 subjects took place. These sessions constituted the independent observations for the non-parametric statistical analysis. Most of the sessions were run in pairs, i.e. 24 subjects were gathered in the lab. The game is repeated over 30 periods and participants are not restricted by choices performed in previous periods. Each period consists of three stages: An institution choice stage (S0), a voluntary contribution stage (S1), and a sanctioning stage (S2). In stage S0 the participants simultaneously and independently choose between a sanctioning institution (SI) and a sanction-free institution (SFI) in which neither positive sanctioning (rewards) nor negative sanctioning (punishment) is possible. In stage S1, each participant is informed about the number of participants in each institution and in case the institution is occupied by at least two participants a public goods game is played with all participants who have chosen the same institution in S0. If only one subject joins an institution the subject’s total endowment is automatically transferred to her/his private account. The public good’s game constitutes a prototypical social dilemma in which each player is endowed with 20 money units (MUs) and may contribute between 0 and 20 MUs to a project which benefits the entire group. Each MU contributed to the public good is deducted from the contributor’s private account and creates a benefit of 1.6 MUs for the entire group. This group benefit is equally distributed among the group members, i.e. if a group consists of n members each member profits by 1.6/n MUs fromGürerk, Irlenbusch & Rockenbach 3 each 1 MU contributed (1.6/n is the marginal per capita return MPCR). If, for example, only one group member contributes the total endowment of 20 and the other n–1 group members contribute nothing, the public good amounts to 20·1.6 and the contributor’s profit is 20·1.6/n while each free-riders’ profit is 20+20·1.6/n. If, however, all n group members contribute an identical amount of x, with 0≤x≤20, the public good is n·x·1.6 and each member achieves a profit of 20–x+1.6·x = 20+0.6·x. Hence for an identical contribution x of all group members the net benefit of each group member is 0.6·x independent from the group size n. The MUs not contributed to the public good are transferred to the participant’s private account. Thus, the provider’s return from one additional MU is less than 1 but the group’s return exceeds 1. Since the cost of providing is higher than the individual return, it is always in the material self-interest of any subject to free-ride on the contributions of the others and to keep all MUs for the private account. If all participants follow their material self-interest, nobody contributes to the public good and each participant achieves a payoff of 20 MUs. Because the group’s return of each MU invested is greater than 1, it is in the collective interest that all group members contribute their entire endowment to the group project. These diametrically opposed individual and collective interests constitute the social dilemma in public good provision. After the players have simultaneously made their contribution decisions, they are informed about the contributions of each member in the own group. At the beginning of stage S2 each player receives additional 20 tokens independent of the affiliation choice in S0. In SFI these tokens are directly transferred to the player’s private account without any decisions required, i.e. sanctioning was not possible. In SI these tokens may be used to positively or negatively sanction other members of SI by assigning between zero and 20 tokens to other members. Each player is free to choose which of the other members of SI she/he wants to positively and/or negatively sanction and to determine the amount of allocated sanctioning tokens to each of those players. She/he is free to allocate different numbers of sanctioning tokens to different individuals with the only restriction thatGürerk, Irlenbusch & Rockenbach 4 the sum of allocated tokens is limited to at most 20. Tokens not used for sanctioning are transferred to the player’s private account. Each token employed as a negative sanction costs the punished member 3 MUs and the punishing member 1 MU. Each token employed as a positive sanction yields the receiving member 1 MU and costs the employing member 1 MU. The leverage in the negative sanctioning mechanism is motivated by the understanding that punishment is more costly for the punished individual than for the punisher. We assume that the leverage in positive sanctioning is smaller and does not create any efficiency gains. The efficiency loss of negative sanctioning as well as the efficiency neutrality of positive sanctioning excludes efficiency gains solely by applying these instruments.