Strategic thinking in public goods games with teams

https://doi.org/10.1016/j.jpubeco.2018.03.007Get rights and content

Highlights

  • We experimentally study team play in public goods games.

  • Teams are compared with individuals under Partners and Strangers matching.

  • Team chat logs reveal how subjects reason about the game.

  • We find concern for future cooperation and limited backward induction.

  • We also find evidence of confusion and explore its sources.

Abstract

We experimentally investigate team behavior in repeated public goods games and use team chat logs to study motives for contribution. Subjects are matched into two-person teams, and each team makes a joint decision in each period. We compare teams with individuals and find similar overall contributions. However, initial contribution is higher and endgame effects are more pronounced for teams. We examine strategic discussions within teams and find strong evidence of concern for repeated game effects and limited backward induction. We also find evidence of confusion and explore its potential sources.

Introduction

Team decision making is widespread in social dilemmas in the field. Families contribute to charities, churches, and neighborhood watch efforts. On a larger scale, firms, non-profit organizations, and governments contribute to disaster relief projects and pollution abatement. We study team behavior in repeated public goods games to address two primary questions. First, we examine whether individuals and teams differ in their contribution decision. Second, we use team chat logs to investigate team contribution motives. While recent studies by Kamei (2016) and Auerswald et al. (2016) have begun examining public goods games with teams, most experiments focus on individuals. Furthermore, studying teams allows us to examine strategic thinking through content analysis of discussions between team members making a joint contribution decision, which was not a focus of these related studies. Strategic discussions within teams provide a direct window into the decision making process. Examining teams can thus yield new insight into motives for contribution.

The method of examining team chat logs has recently been used to gain valuable insight into strategic thinking in other contexts such as the prisoner's dilemma (Kagel and McGee, 2016, Cason et al., 2017, Cason and Mui, 2017), signaling games (Cooper and Kagel, 2005), legislative bargaining (Bradfield and Kagel, 2015), ultimatum bargaining (Arkes et al., 2015), coordinated resistance games (Cason and Mui, 2015), and beauty contest games (Burchardi and Penczynski, 2014, Penczynski, 2016). In our setup, groups are composed of multiple decision makers, and each decision maker in a group is a team of two subjects who make a single joint decision in each period. Team members communicate with one another via text chat. Importantly, communication with other teams is not permitted, and team members have identical payoffs so their incentives are aligned, motivating them to work together to form a profitable strategy.1 By examining chat logs among team members, it is possible to better understand how subjects reason about the public goods game, and what concerns motivate their contribution decisions.

This “two heads” team chat method may be thought of as an elicitation procedure, similar in principle to various methods of eliciting beliefs, preferences, or strategies in economic experiments. However, the data being elicited are qualitative chat messages that reveal thought processes leading to contribution decisions. Chat messages are coded by research assistants into one or more of several categories such as discussing pro-social preferences, payoff maximization, and repeated game effects. These chat codes are used to learn about the underlying motives and strategic thinking of subjects in the experiment.

To examine the effect of playing in teams on contribution, we compare team behavior in public goods games to a baseline individual treatment. To investigate strategic thinking about repeated game effects and backward induction, we also compare the cases of random Strangers matching and fixed Partners matching of decision makers into groups. By comparing team chat logs in Partners and Strangers treatments, our experiment yields new insights on how motivations for contribution vary with repeated interaction in fixed groups. In this way, our experiment is related to the literature comparing Partners and Strangers in public goods games (Andreoni, 1988, Croson, 1996, Keser and van Winden, 2000, Andreoni and Croson, 2008) as well as the much broader literature on cooperation and free-riding in social dilemmas (Ledyard, 1995, Ostrom, 2000, Chaudhuri, 2011). To the best of our knowledge, this study is the first to use content analysis of team chat logs to compare strategic thinking with and without repeated interaction in fixed groups in any game.

We find that behavior is largely similar between individuals and teams. However, initial contribution is higher for teams. Furthermore, endgame effects are more pronounced for teams, as free-riding rates in the last period are greater for teams than individuals.

Aggregate contribution is similar for Partners and Strangers. However, we find differences in strategic thinking. Compared to Strangers, Partners more frequently discuss encouraging cooperation in future periods, as well as expectations of others' future choices. Nonetheless, concern for repeated game effects occurs with Strangers matching as well.

We find evidence of limited backward induction, with discussions of endgame effects mostly contained in the last few periods. Moreover, discussion of higher-order beliefs integral to the backward-induction process is very rare. Team discussions also reveal evidence of confusion, but relatively little direct evidence of pro-social preferences. Finally, we explore sources of possible confusion revealed in the chat logs, and discuss potential methodological implications for the design of future experiments.

Section snippets

Related literature

Many studies examine motivations for voluntary contribution to public goods and cooperation in social dilemmas more generally (Ledyard, 1995, Chaudhuri, 2011). Various forms of social preferences such as altruism and warm glow have been used to explain voluntary contribution (see, e.g. Andreoni, 1990, Goeree et al., 2002, Crumpler and Grossman, 2008. Several studies on repeated game effects compare fixed Partners matching with random Strangers matching, finding mixed results (Andreoni, 1988,

Experimental design and procedures

We will refer to a decision-making unit (an individual or a team) as a decision maker. In Individual treatments, a decision maker is an individual human subject. In Teams treatments, a single decision maker is a team, composed of two human subjects who make a single joint decision in each period. Groups are composed of three decision makers in all treatments. In all treatments, matching was random, anonymous, and computerized. Partitions between lab stations were used to ensure anonymity.

We

Results

We first examine treatment differences in contribution levels and proportion of free riders. We then analyze the chat logs of team members. The treatments are identified by Individuals or Teams and the matching mechanism, Partners or Strangers. For brevity in the discussion of the results, treatments will be referred to by the acronyms listed in Table 1. For example, the Team-Strangers treatment will be referred to as TS.

Confusion in team chat logs

Previous studies such as Andreoni, 1995, Houser and Kurzban, 2002, and Shapiro (2009) show evidence that confusion plays a substantial role in public goods games. Confusion is typically identified as a residual source of contribution after other explanations such as social preferences are removed. By exploring our team chat logs, it is possible to gain a more direct understanding of the potential sources of confusion and how they might be mitigated. The chat codes assigned by research

Conclusion

In this experiment, we compare contribution choices of individuals and two-person teams in public goods games and examine the motives of teams by analyzing their chat logs. We find similar overall contribution for teams and individuals. However, initial contribution is higher for teams, and there are stronger endgame effects for teams, with more frequent free-riding in the final period. These features of the data are qualitatively similar to the results found in Auerswald et al. (2016), though

Acknowledgments

We thank Jared Weber, Ashley Halloway, Clarence Castro, Jonathan Howard, and Reed Muehler for excellent research assistance. We are grateful to Co-Editor Tim Cason and two anonymous referees for very helpful comments and suggestions. We also thank John Kagel, David Cooper, James Walker, Arlington Williams, Lata Gangadharan, Oleg Korenok, Tim Salmon, Peter McGee, Glenn Dutcher, Kenju Kamei, Edward Millner, Javier Portillo, Andrzej Baranski, David Kingsley, Lawrence De Geest, David Bruner, Dave

References (85)

  • H. Crumpler et al.

    An experimental test of warm glow giving

    J. Public Econ.

    (2008)
  • D.D. Davis et al.

    Group vs. individual performance in a price-searching experiment

    Organ. Behav. Hum. Decis. Process.

    (1996)
  • L. Gangadharan et al.

    Normative conflict and the limits of self-governance in heterogeneous populations

    Eur. Econ. Rev.

    (2017)
  • J.K. Goeree et al.

    Private costs and public benefits: unraveling the effects of altruism and noisy behavior

    J. Public Econ.

    (2002)
  • S.P. Hargreaves-Heap et al.

    ‘Doggedness' or ‘disengagement’? An experiment on the effect of inequality in endowment on behaviour in team competitions

    J. Econ. Behav. Organ.

    (2015)
  • C.A. Insko et al.

    Individual-group discontinuity: the role of a consensus rule

    J. Exp. Soc. Psychol.

    (1988)
  • D.M. Kreps et al.

    Rational cooperation in the finitely repeated prisoners' dilemma

    J. Econ. Theory

    (1982)
  • T. Kugler et al.

    Trust between individuals and groups: groups are less trusting than individuals but just as trustworthy

    J. Econ. Psychol.

    (2007)
  • P.M. Morgan et al.

    Group vs individual performance in mixed-motive situations: exploring an inconsistency

    Organ. Behav. Hum. Decis. Process.

    (2002)
  • S.P. Penczynski

    Persuasion: an experimental study of team decision making

    J. Econ. Psychol.

    (2016)
  • R. Selten et al.

    End behavior in sequences of finite repeated prisoner's dilemma supergames: a learning theory approach

    J. Econ. Behav. Organ.

    (1986)
  • M. Sutter et al.

    Strategic sophistication of individuals and teams. experimental evidence

    Eur. Econ. Rev.

    (2013, November)
  • T.K. Ahn et al.

    Cooperation in PD games: fear, greed, and history of play

    Public Choice

    (2001)
  • J. Andreoni

    Impure altruism and donations to public goods: a theory of warm-glow giving

    Econ. J.

    (1990)
  • J. Andreoni

    Cooperation in public goods experiments: kindness or confusion?

    Am. Econ. Rev.

    (1995)
  • J. Andreoni et al.

    Rational cooperation in the finitely repeated prisoners' dilemma: experimental evidence

    Econ. J.

    (1993)
  • H. Arkes et al.

    Effects of a management — labor context and team play on ultimatum game outcome

    Working paper

    (2015)
  • R. Ashley et al.

    Motives for giving: a reanalysis of two classic public goods experiments

    South. Econ. J.

    (2010)
  • H. Auerswald et al.

    Teams contribute more and punish less

    Working Paper

    (2016)
  • G. Bornstein et al.

    Individual and group behavior in the ultimatum game: are groups more ‘rational’ players?

    Exp. Econ.

    (1998)
  • T.N. Cason et al.

    Prior interaction, identity, and cooperation in the inter-group prisoner's dilemma

    Working paper

    (2017)
  • T.N. Cason et al.

    A laboratory study of group polarisation in the team dictator game

    Econ. J.

    (1997)
  • T.N. Cason et al.

    Individual versus group play in the repeated coordinated resistance game

    J. Exp. Pol. Sci.

    (2015)
  • T.N. Cason et al.

    Individual versus group choices of repeated game strategies: a strategy method approach

    Working paper

    (2017)
  • G. Charness et al.

    Understanding social preferences with simple tests

    Q. J. Econ.

    (2002)
  • G. Charness et al.

    Groups make better self-interested decisions

    J. Econ. Perspect.

    (2012)
  • A. Chaudhuri

    Sustaining cooperation in laboratory public goods experiments: a selective survey of the literature

    Exp. Econ.

    (2011)
  • C.C. Clogg et al.

    Statistical methods for comparing regression coefficients between models

    Am. J. Sociol.

    (1995)
  • J. Cohen

    A coefficient of agreement for nominal scales

    Educ. Psychol. Meas.

    (1960)
  • D. Cooper et al.

    Are two heads better than one? Team versus individual play in signaling games

    Am. Econ. Rev.

    (2005)
  • C. Cox et al.

    Framing and feedback in social dilemmas with partners and strangers

    Games

    (2015)
  • C.A. Cox et al.

    Revealed reputations in the finitely-repeated prisoners' dilemma

    Econ. Theory

    (2015)
  • Cited by (0)

    View full text