Alleviation and Sanctions in Social Dilemma Games

Games 2015, 6, 368-380; doi:10.3390/g6030368 OPEN ACCESS games ISSN 2073-4336 www.mdpi.com/journal/games Article Alleviation and Sanctions in Social Dilemma Games Michalis Drouvelis Department of Economics, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK; E-Mail: ; Tel.: +44-(0)-121-4146040 Academic Editor: Ananish Chaudhuri Received: 29 June 2015 / Accepted: 16 September 2015 / Published: 21 September 2015 Abstract: This paper reports an experiment which compares behaviour in two punishment regimes: (i) a standard public goods game with punishment in which subjects are given the opportunity to punish other group members (democratic punishment regime) and (ii) a public goods game environment where all group members exogenously experience an automatic reduction of their income (irrespective of their behaviour) and are given the opportunity to alleviate the automatic penalty (undemocratic punishment regime). We employ a within-subjects design where subjects experience both environments and control for order effects by alternating their sequence. Our findings indicate that average contributions and earnings in the undemocratic punishment environment are significantly lower relative to the standard public goods game with punishment. We also observe that in the undemocratic environment average contributions decay over time only when subjects have experienced the standard public goods game with punishment. As a result, alleviation is significantly less when subjects have experienced the standard public goods game with punishment compared to when they do not have such experience. However, the assignment of punishment is robust irrespective of the order in which the games are played. Keywords: cooperation; reciprocity; alleviation; sanctions; public good experiments JEL Codes: C9; D7; H4 1. Introduction A central theme in the behavioural sciences is the examination of the ability of punishment regimes, where sanctioning is assigned endogenously by group members to their peers, to regulate self-interested behaviour in social dilemma games (as in [1,2]). We refer to these environments as Games 2015, 6 369 “democratic punishment regimes”. Numerous laboratory experiments on public goods have shown that such democratic punishment regimes foster high cooperation rates (for recent overviews, see [3,4]). However, this cooperation enhancing effect has been found to be sensitive to a number of factors, such as low effectiveness of punishment (see [5–8]), second-round punishment opportunities (see [9–11]), and antisocial punishment (see [4,12,13]). The evidence from these experiments suggests that, under certain conditions, peer punishment can have little or negative effect on cooperation. The aim of this paper is to explore experimentally how subjects respond in a democratic punishment regime when they have experience with an undemocratic punishment regime and vice versa. To generate an undemocratic punishment regime, we propose a variant of the standard punishment game, which we refer to as the “default punishment game”. In this game, group members participate in a two-stage game. In the first stage, they are engaged in a standard linear public goods game, in which they have to decide how much of their initial endowment they are willing to contribute to the public good (see [14]). The game is parameterized so that it captures the tension between personal interests and social benefits. At the beginning of the second stage, we introduce a new element: each group member’s income is exogenously reduced by a fixed automatic penalty imposed on them—the default punishment. This implies that all group members unconditionally receive a decrease in their monetary income, irrespective of their first stage behaviour. During the second stage, after contribution decisions have been anonymously revealed, subjects are given the opportunity to alleviate the exogenous default punishment of others at some cost to themselves. Our environment has the interesting feature that the fixed income reduction does not depend on the individual behaviour. The automatic penalty is undemocratic in the sense that it is exogenously implemented and the sanctioning mechanism is not being any more social as it cannot identify individual defectors or cooperators.1 We thus assume that our punishment regime generates a less democratic environment in relation to the standard punishment game where punishment is meted out endogenously by group members. Our paper is also related to the literature examining behaviour in situations where subjects are given the possibility (typically using voting procedures) to choose the public goods game environment they would like to interact in, as well as the punishment rules governing these environments (e.g., [16–19]). 2 In sum, this literature suggests that letting subjects democratically choose which regime they prefer positively affects cooperative behaviour. In our paper, the automatic penalty is exogenously assigned to all group members (including those with cooperative attitudes) and it is in this respect that our punishment regime is undemocratic. By having subjects experience an undemocratic punishment environment before they play the standard punishment game, we can assess whether our undemocratic environment impacts on individuals’ 1 2 The effects of blind punishment on public good provision have been recently studied by [15]. In particular, they use a punishment mechanism based on random exclusions. Yet, their design is distinctive to ours in two main respects concerning the rules governing the implementation of punishment. First, their punishment scheme is still social and pursues a collective goal. Good teams were never punished under their sanctioning system. Second, unlike our experiment where subjects are given the opportunity to alleviate the automatic penalty, in their design there is no second stage in which individuals can correct the unfairness of the blind punishment. Their findings suggest that random exclusions generate more public good provision (compared to a standard public good game without punishment) and promote efficiency in a significant way. For a review of the experimental literature on the workings of democratic institutions, see [20]. Games 2015, 6 370 expectations of how punishment works. We also reverse the order of the two games. This allows us to test whether, and if so how, experience of a democratic punishment regime affects behaviour in the undemocratic punishment regime. Our findings indicate that more pronounced detrimental effects on behaviour are present in the undemocratic punishment regime when subjects have experienced the democratic punishment regime. However, this is not the case for the democratic punishment regime where behaviour is not affected by the experience (or not) of the undemocratic default punishment game. Overall, we find that performance in the undemocratic punishment regime is worse in relation to th (...truncated)