Skip to main content

Delegation and Public Pressure in a Threshold Public Goods Game


Many public goods cannot be provided directly by interested parties (e.g. citizens), as they entail decision-making at nested hierarchical scales: at a lower level individuals elect a representative, while at a higher scale elected delegates decide on the provision level, with some degree of scrutiny from their constituency. Furthermore, many such decisions involve uncertainty about the magnitude of the contribution that is needed for the good to be provided (or bad to be avoided). In such circumstances delegates can serve as important vehicles for coordination by aggregating societal preferences for provision. Yet, the role of delegation in threshold public goods games is understudied. We contrast the behavior of delegates to that of self-representing individuals in the avoidance of a public bad in an experimental setting. We randomly assign twelve subjects into four teams and ask each team to elect a delegate via majority voting. The elected delegates play several variants of a one-shot public goods game in which losses can ensue if the sum of their contributions falls short of a threshold. We find that when delegation is coupled with a mild form of public pressure, it has a significantly negative effect on contributions, even though the non-delegates can only signal their preferred levels of public good contributions. The reason is that delegates give more weight to the least cooperative suggestion: they focus on the lower of the two public good contributions recommended by their teammates.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2


  1. 1.

    Throughout this paper, we refer to a “group” as a society that consists of “teams.” Under delegation, each team elects a “delegate”, who acts on behalf of the team and shares the earnings equally with her team members.

  2. 2.

    In linear public goods games the dominant strategy is not investing in the public good at all, while the threshold public goods game we use, due to the non-linearity introduced by the threshold, entails different strategies depending on what one assumes about the other delegates’ contributions. If the threshold is perceived to be out of reach, we are back in the above case; if instead the threshold is within reach, provision choices can either be strategic substitutes (if the decision-maker does not consider herself pivotal for reaching the target investment) or complements otherwise.

  3. 3.

    A second reason for the importance of delegation in our threshold public goods experiment is that, as argued in footnote 2, and unlike in the linear setups employed by Hamman, Weber, and Woon, 2011 as well as Kocher, Tan and Yu, 2018, provision choices can be strategic substitutes. This implies that the constituency may have an incentive to strategically delegate to agents who have a lower valuation for the public good than they have themselves, in order to free-ride on the others’ provision (Habla and Winkler 2016). While testing for strategic delegation is not the goal here, it is interesting to see whether we observe such pattern in the election of delegates.

  4. 4.

    Note that the threshold public goods game in our experiment encompasses both rounds of play where coordination on threshold provision is individually optimal, as well as rounds in which the typical cooperation dilemma arises (i.e. one is better off not contributing). However, our design is not necessarily representative of other types of strategic interactions, such as those occurring when individuals access common property resources characterized by rivalry in extraction. See Stoop et al. (2012) for a comparison of recreational fishermen’s behaviour in the lab and in the field.

  5. 5.

    Put differently, delegation has the potential to bring groups closer to the non-cooperative equilibrium (where the threshold public good is not provided), due the negative effect of messages on delegate contributions, since the latter are largely driven by the lowest value suggested by the constituency.

  6. 6.

    This game resembles the stag-hunt game in which players have to decide between socially cooperative and safe strategies. By contributing positive amounts, a team runs the risk of becoming the “sucker” of the group, which lowers the attractiveness of the SPC.

  7. 7.

    Sogang University did not yet have an experimental laboratory at the time we ran our experiment, so we devised a simple paper and pencil protocol which worked seamlessly, and whose implementation was facilitated by RAs and a text messaging application that ensured that the treatments took place in a controlled environment. Furthermore, group identity was important for treatments 2 and especially 3, as explained below, and our design is arguably more effective for this purpose than a standard computer lab.

  8. 8.

    To keep the earnings and, thus, the monetary incentives unchanged between treatments, we set the exchange rate between the laboratory dollar and the USD in the first treatment at 3 laboratory dollars = 1 USD. In the other treatments, the exchange rate is 1 laboratory dollar = 1 USD. Thus, in all treatment decision makers (self-representing individuals or delegates) allocate 120 laboratory dollars.

  9. 9.

    We asked several control questions before the practice phase took place, to make sure that all subjects understood the game. We did not start the experiment until every subject had answered correctly all questions. The instructions, control questions, decision sheets, and survey questions can be found at

  10. 10.

    We are going to control the order of play in empirical analysis. Similarly, we are going to discuss the impact of the specific game played in practice phase on decision makers as it might produce more non-cooperative behaviour in Sect. 4.

  11. 11.

    Details about the ID cards and other experimental procedures are in Online Appendix, Sect. 4.

  12. 12.

    For the summary of treatments, see Online Appendix, Table 1.

  13. 13.

    The average earnings were 10,195 KRW, with a maximum of 40,000 and a minimum of 1000. Additionally, subjects receive the show-up fee of 10,000 KRW (about 10 USD).

  14. 14.

    Furthermore, as we explain in Sect. 3, we rule out strategic communication between teams.

  15. 15.

    See Online Appendix, Table 2.

  16. 16.

    Further results, including contributions in each group, pledges and survey findings are contained in Online Appendix, Sect. 6.

  17. 17.

    Note that Mijr cannot be included for noD since there is no delegation in this treatment. Non-delegates’ opinions are not delivered to delegates in DnoM. Therefore, there should be no effect of non-delegates’ opinions on delegates’ decisions in DnoM. This will serve as a validity test of our experimental design.

  18. 18.

    For round-specific fixed effects, although there are six rounds, we can only include three dummies owing to perfect linear collinearity with p75r, p95r, and UTr.

  19. 19.

    The full results are presented in Online Appendix, Table 3.

  20. 20.

    Furthermore, the results about the age effect are intriguing. In noD, age does not matter, while it has a positive effect in DnoM, where older delegates contribute more. However, the effect is opposite in DM; older delegates contribute less after controlling for non-delegates’ opinions. While interesting, one should be cautious about inferring from these results, given that there is little age variation among the subjects.

  21. 21.

    Note that this result is robust to the exclusion of rounds 2 and 5, where the payoff from SPC is not strictly greater than the payoff from NC, and hence it is rational to follow the lower message (see Table 1). That is, the results of Table 3 hold after excluding rounds 2 and 5: average message and lower/higher messages do not matter in DnoM in columns (3) and (4), but they are significant in DM in columns (6) and (7). Note also that the model in column (8) restricts the sample to round 1 only, and confirms the result in the “cleanest” setting (devoid of this and other confounders such as contamination across rounds).

  22. 22.

    We examine both delegates’ contributions and non-delegates’ messages for all 192 choices (32 delegates × 6 treatments) in DM. Of all 192 choices, 55% are closer to the lower opinion, while only 21% are closer to the higher opinion. See Online Appendix, Fig. 1.

  23. 23.

    For the full results, see Online Appendix, Table 4. More specifically, we used the specification of column (7) in Table 3 for various subsamples. For example, there are 192 observations in DM, and 126 are male delegates and 66 are female delegates. For other subsamples, we use information collected by post-experiment surveys (the first four subsamples), by experimental design (single certain threshold or double uncertain thresholds), or by information on pledges and expectations (the last four subsamples).

  24. 24.

    Recall that, unlike in DnoM, non-delegates in DM can signal their preferred contributions to their delegates. However the results in Table 5 show that non-delegates in DM did not behave differently from those in DnoM. Indeed, the last column shows that we cannot reject the null that non-delegates in DM are the same as those in DnoM. This means that the lower public good contributions by the delegates in DM are attributable to the fact that they responded to the messages sent by their constituency.

  25. 25.

    Note that, due to the experimental design, we cannot attribute the difference between the DM and DnoM treatments to messages alone. The reason is that while non-delegates were also present in the decision room in DM, they sat in a different room in DnoM. For this reason, we can only conclude that public pressure from the constituency, in the form of a combination of messages and passive auditing in the decision room, induces the delegates to decrease group contributions to the public good.

  26. 26.

    See Bleichrodt and Wakker (2015) for more about this evidence and applications of regret theory.


  1. Barrett S, Dannenberg A (2012) Climate negotiations under scientific uncertainty. Proc Natl Acad Sci 109(43):17372–17376

    Article  Google Scholar 

  2. Bleichrodt H, Wakker PP (2015) Regret theory: a bold alternative to the alternatives. Econ J 125:493–532

    Article  Google Scholar 

  3. Bolton G, Ockenfels A, Stauf J (2015) Social responsibility promotes conservative risk behavior. Eur Econ Rev 74:109–127

    Article  Google Scholar 

  4. Cadsby CB, Maynes E (1999) Voluntary provision of threshold public goods with continuous contributions: experimental evidence. J Public Econ 71(1):53–73

    Article  Google Scholar 

  5. Charness G, Jackson MO (2009) The role of responsibility in strategic risk-taking. J Econ Behav Org 69:241–247

    Article  Google Scholar 

  6. Charness G, Rigotti L, Rustichini A (2007) Individual behavior and group membership. Am Econ Rev 97(4):1340–1352

    Article  Google Scholar 

  7. Croson R, Marks M (2000) Step returns in threshold public goods: a meta-analysis and experimental analysis. Exp Econ 2(3):239–259

    Article  Google Scholar 

  8. Dannenberg A, Löschel A, Paolacci G, Reif Christiane, Tavoni Alessandro (2015) On the provision of public goods with probabilistic and ambiguous thresholds. Environ Resour Econ 61:365–383

    Article  Google Scholar 

  9. Habla W, Winkler R (2016) Strategic delegation and international permit markets: why linking may fail. EfD discussion paper series, pp 16–12

  10. Hamman JR, Weber RA, Woon J (2011) An experimental investigation of electoral delegation and the provision of public goods. Am J Political Sci 55(4):738–752

    Article  Google Scholar 

  11. İriş Doruk (2018) Representation and social regret-aversion in risk-taking. Korean J Ind Org 26(3):1–17

    Google Scholar 

  12. Kocher MG, Tan F, Yu J (2018) Providing global public goods: electoral delegation and cooperation. Econ Inquiry 56(1):381–398

    Article  Google Scholar 

  13. Loomes G, Sugden R (1982) Regret theory: an alternative theory of rational choice under uncertainty. Econ J 92(368):805–824

    Article  Google Scholar 

  14. Milinski M, Sommerfeld RD, Krambeck H-J, Reed FA, Marotzke J (2008) The collective-risk social dilemma and the prevention of simulated dangerous climate change. Proc Natl Acad Sci USA 105(7):2291–2294

    Article  Google Scholar 

  15. Song F (2008) Trust and reciprocity behavior and behavioral forecasts: individual versus group-representatives. Games Econ Behav 62:675–696

    Article  Google Scholar 

  16. Stoop J, Noussair CN, van Soest D (2012) From the lab to the field: cooperation among fishermen. J Political Econ 120(6):1027–1056

    Article  Google Scholar 

  17. Tavoni A, Dannenberg A, Kallis G, Löschel A (2011) Inequality, communication and the avoidance of disastrous climate change in a public goods game. Proc Natl Acad Sci USA 108:11825–11829

    Article  Google Scholar 

  18. Zeelenberg M (1999) Anticipated regret, expected feedback and behavioral decision making. J Behav Dec Making 12(1):93–106

    Article  Google Scholar 

Download references


We would like to thank Astid Dannenberg, Michael Finus, Andreas Lange, Tatsuyoshi Saijo, Timo Goeschl (editor), and two unknown referees for their helpful suggestions. All remaining errors are ours. This work was supported in part by funding from the National Research Foundation of Korea (Grant No. 201322021.01). Lee’s work was supported by Research Resettlement Fund for the new faculty of Seoul National University.

Author information



Corresponding author

Correspondence to Doruk İriş.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

İriş, D., Lee, J. & Tavoni, A. Delegation and Public Pressure in a Threshold Public Goods Game. Environ Resource Econ 74, 1331–1353 (2019).

Download citation


  • Delegation
  • Cooperation
  • Threshold public goods game
  • Climate experiment

JEL Classification

  • C72
  • C92
  • D81
  • H4
  • Q54