Communication & competition


Charness and Dufwenberg (Am. Econ. Rev. 101(4):1211–1237, 2011) have recently demonstrated that cheap-talk communication raises efficiency in bilateral contracting situations with adverse selection. We replicate their main finding and extend their design to include competition between agents. We find that communication and competition act as “substitutes:” communication raises efficiency in the absence of competition but not with competition, and competition raises efficiency without communication but lowers efficiency with communication. We briefly review some behavioral theories that have been proposed in this context and show that each can explain some but not all features of the observed data patterns. Our findings highlight the fragility of cheap-talk communication and may serve as a guide to refine existing behavioral theories.

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

Fig. 1
Fig. 2
Fig. 3


  1. 1.

    We doubled the payoffs in Charness and Dufwenberg (2011) to make the monetary incentives more salient.

  2. 2.

    Choosing “In” yields an expected payoff of only 1/3×5/6×24=20/3 for the principal.

  3. 3.

    Stigler (1987, p. 531) defines competition as “a rivalry between individuals … that arises whenever two or more parties strive for something that all cannot obtain.” Treatment “3C-WDR” captures this definition while keeping the incentives for the principal and the selected agent the same as in the “2C-WDR” treatment. This allows us to isolate the effect of competition when communication is possible. Furthermore, the comparison between treatments “3C-WDR” and “3NC-WDR” allows us to measure the effect of communication in the presence of competition.

  4. 4.

    One difference is that our experiments were computerized using zTree (Fischbacher 2007). In all treatments with communication, subjects could choose to remain silent by simply clicking the continue button.

  5. 5.

    For instance, agents’ types were randomly determined by the program and the fraction of high-type agents varied from 28.6 % to 41.7 % across treatments. To correct for this variability, the predicted fraction of efficient outcomes, \(p^{In}({1 \over3}+{2 \over3}p^{DR})\), uses the ex ante probabilities for each type. Here p In denotes the principal’s “In” rate and p DR the low-agent’s “Don’t Roll” rate. In the communication treatments, the “In” and “Don’t Roll” rates may depend on the agent’s message, m, which, in turn, may depend on the agent’s type. The predicted fraction of efficient outcomes now becomes \(\sum_{m} p^{In}(m)({1 \over3}P_{H}(m)+{2 \over3}P_{L}(m)p^{DR}(m))\) where P L (m) and P H (m) are the probabilities that a low-type or high-type agent sends message m respectively. See Sect. 3.2 for a detailed discussion on how messages were classified and how the fraction of efficient outcomes were calculated for the case with agent competition.

  6. 6.

    More specifically, a two-sided proportion test shows no significant difference at the 10 % level between the “In” rates in “NDR” vs “WDR”, “NDR” vs “C&D”, “WDR” vs “C&D”, and “pooled” vs “C&D”, for both the “NC” and “C” treatments respectively. The same no-difference result holds for the “Don’t Roll” rate and the percentage of efficient outcomes in both the “NC” and “C” treatments respectively. All p-values reported in this paper are two-sided, unless otherwise stated.

  7. 7.

    Efficiency rises from 30.1 % in 2NC to 53.5 % in 3NC (p=0.026) but it falls from 64.4 % in 2C to 37.5 % in 3C (p=0.014).

  8. 8.

    Efficiency rises from 30.1 % in 2NC to 64.4 % in 2C (p<0.001) but it falls from 53.5 % in 3NC to 37.5 % in 3C (p=0.162). Had we based our null hypotheses on Charness and Dufwenberg’s (2011) finding that communication raises efficiency then this null hypothesis would be rejected in a one-sided test (p=0.081). Note that Charness and Dufwenberg (2011) use a one-sided test to evaluate the effect of communication on efficiency.

  9. 9.

    As in Charness and Dufwenberg (2011) most, but not all, messages can be captured with this coding scheme. Other types of messages are “PL” when a low-type agent only discloses her type with no promise about the action she will take, “PR” when the agent only promises to “Roll” without disclosing her type, “PH” when the agent claims to be of high type with no promise about the action, and “DR” when the agent promises to choose “Don’t Roll” without disclosing her type. The first two messages were classified as “NP,” the third message as “HR” and the fourth message as “LD.” Finally, empty talk messages and no messages are included in “NP.”

  10. 10.

    For example, in 2C (pooled) sample, the principal chose “In” in 6/7 cases when the low-type agent sent an “HR” message and in 12/13 cases when the high-type agent sent an “HR” message.

  11. 11.

    The Fisher exact test comparing the NP, LD and HR messages sent by low-type and high-type agents yields p<0.001 for treatment “2C” and p=0.010 for treatment “3C.”

  12. 12.

    For low-type agents the difference is close to being significant with p=0.124, for high-type agents p=0.073, and for the pooled messages p=0.041 using the Fisher exact test.

  13. 13.

    The proportion of “NP” messages sent by the low-type and high-type agents increases from 27 % to 45 % and from 7 % to 38 % respectively. A two-sided proportion test shows that these increases are significant (p=0.09 for low-type agents and p=0.04 for high-type agents).

  14. 14.

    In treatment 3C, out of 74 messages there were 32 NP messages that can be broken down as follow: 11 silent, 2 PR messages, 3 PL messages, and 16 empty talk messages. In 2C, out of 47 messages there were 10 NP messages: 1 silent, 3 PL messages, and 6 empty talk messages. A Fisher exact test cannot reject the null hypothesis that the distributions of message types within the NP category are the same between 2C and 3C (p=0.187). However, there are significantly more NP messages in 3C (p=0.013).

  15. 15.

    When a low-type agent sends message m, predicted efficiency is \(P_{L}^{matched}(m)p^{In}(m)p^{DR}(m)\) and when a high-type agent sends message m it is \(P_{H}^{matched}(m)p^{In}(m)\).

  16. 16.

    For example, the second entry in the top row indicates that 20 % of the time the principal selects the “NP” message from the pair (“NP”,“LD”). The first entry in the second row shows the “LD” message is selected from such a pair with complementary probability. More generally, the sum of the selection matrix and its transpose yields 1 in all entries since one of the two messages is selected. For the same reason the diagonal elements are 1/2.

  17. 17.

    For the “NP” and “LD” messages these differences are significant (p=0.04 and p=0.03 respectively).

  18. 18.

    The conditional “Don’t Roll” rates are 100 % (5/5) and 87 % (13/15) in treatments with and without competition respectively. The difference is not significant (p=0.389).

  19. 19.

    The percentages of lies are 37.5 % (9/24) and 27.3 % (9/33) in treatments with and without competition respectively. The difference is not significant (p=0.412).

  20. 20.

    An interesting extension is to let the principal’s “In” rate depend on both messages received. In this case, the predicted fraction of efficient outcomes drops to 34.2 % and the difference between “2C” and “3C” is significant at the 5 % level (p=0.0475) and the difference between “3C” and “3NC” is significant at the 10 % level (p=0.09).

  21. 21.

    See, for instance, Gneezy (2005), Vanberg (2008), Ellingsen et al. (2009), and Sutter (2009). Charness and Dufwenberg (2010) and Serra-Garcia et al. (2011) study how the results depend on the type of language used by comparing bare versus rich messages and vague versus precise messages respectively.

  22. 22.

    For models of lie aversion see Ellingsen and Johannesson (2004), Demichelis and Weibull (2008), Vanberg (2008), and Kartik (2009). To model guilt, two notions are offered by Battigalli and Dufwenberg (2007, 2009): simple guilt and guilt-from-blame. Charness and Dufwenberg (2006) and Ellingsen et al. (2010) provide evidence of simple guilt in trust games while Charness and Dufwenberg (2011) test guilt-from-blame.

  23. 23.

    In repeated trust games that allow for reputation building, Huck et al. (2012) find that competition among trustees significantly improves trust and trustworthiness and, hence, efficiency. In contrast, Fehr et al. (1998) and Brandts and Charness (2004) find that competition does not significantly alter behavior in repeated gift-exchange games. In a one-shot trust game where reputation formation is not possible, Bauernschuster et al. (2012) find that when trustees can select from multiple trustors, the trustor with the highest offered amount is always chosen but is returned a significantly lower amount than trustors receive in a control treatment without competition. Roth et al. (1991) and Grosskopf (2003) investigate how competition affects bargaining outcomes.

  24. 24.

    See Fehr and Schmidt (1999), Bolton and Ockenfels (2000), Rabin (1993), Dufwenberg and Kirchsteiger (2004), and Falk and Fischbacher (2006).

  25. 25.

    Rode (2010) is the only paper we are aware of that studies the interaction between communication and competition. In Rode’s experiment, pairs of subjects either play a cooperative coordination game or a competitive matching pennies game before being matched with a different opponent in a cheap-talk sender-receiver game. Rode finds that the competitive nature of the initial game does not increase the number of lies but it does decrease trust as it leads subjects to believe that the cheap-talk game is a situation of conflicting interest.

  26. 26.

    For the payoffs of Fig. 1, it is trivial to verify that this is an equilibrium when the cost of lying k≥6.

  27. 27.

    Indeed, if the principal believes that the agent will choose “Don’t Roll” with probability one then her “In” choice is the unique Pareto efficient action, which entails zero kindness. As a result, the low-type agent has no incentive to keep the promise.

  28. 28.

    Communication can be efficiency improving with more than two people if they have a common objective as is the case, for instance, with jury decision making (Goeree and Yariv 2011).

  29. 29.

    Goeree and Zhang (2013a) documents significant efficiency-enhancing effect of communication with or without competition.

  30. 30.

    Preliminary evidence suggests that cheap-talk works well in bilateral bargaining but not in markets with a larger number of traders (Goeree and Zhang 2013b).

  31. 31.

    Zhang (2013) reports that it is the larger group that benefits from within-group communication at the expense of the smaller group in participation games.


  1. Battigalli, P., & Dufwenberg, M. (2007). Guilt in games. American Economic Review Papers and Proceedings, 97(2), 170–176.

    Article  Google Scholar 

  2. Battigalli, P., & Dufwenberg, M. (2009). Dynamic psychological games. Journal of Economic Theory, 141(1), 1–35.

    Article  Google Scholar 

  3. Bolton, P., & Dewatripont, M. (2005). Contract theory. Cambridge: MIT Press.

    Google Scholar 

  4. Bolton, G., & Ockenfels, A. (2000). ERC: a theory of equity, reciprocity, and competition. The American Economic Review, 90(1), 166–193.

    Article  Google Scholar 

  5. Brandts, J., & Charness, G. (2004). Do labour market conditions affect gift exchange? Some experimental evidence. Economic Journal, 114, 684–708.

    Article  Google Scholar 

  6. Brandts, J., & Cooper, D. (2007). It’s what you say not what you pay: an experimental study of manager-employee relationships in overcoming coordination failure. Journal of the European Economic Association, 5(6), 1223–1268.

    Article  Google Scholar 

  7. Bauernschuster, S., Falck, O., & Große, N. (2012). When trustors compete for the favour of a trustee—a laboratory experiment. Journal of Economic Psychology, 34(1), 133–147. 2013.

    Google Scholar 

  8. Charness, G., & Dufwenberg, M. (2006). Promises and partnership. Econometrica, 74(6), 1579–1601.

    Article  Google Scholar 

  9. Charness, G., & Dufwenberg, M. (2010). Bare promises: an experiment. Economics Letters, 107, 281–283.

    Article  Google Scholar 

  10. Charness, G., & Dufwenberg, M. (2011). Participation. The American Economic Review, 101(4), 1211–1237.

    Article  Google Scholar 

  11. Demichelis, S., & Weibull, J. (2008). Language, meaning, and games: a model of communication, coordination, and evolution. The American Economic Review, 98(4), 1292–1311.

    Article  Google Scholar 

  12. Dufwenberg, M., & Kirchsteiger, G. (2004). A theory of sequential reciprocity. Games and Economic Behavior, 47(2), 268–298.

    Article  Google Scholar 

  13. Ellingsen, T., & Johannesson, M. (2004). Promises, threats, and fairness. Economic Journal, 114, 397–420.

    Article  Google Scholar 

  14. Ellingsen, T., Johannesson, M., Lilja, J., & Zetterqvist, H. (2009). Trust and truth. Economic Journal, 119, 252–276.

    Article  Google Scholar 

  15. Ellingsen, T., Johannesson, M., Tjøtta, S., & Torsvik, G. (2010). Testing guilt aversion. Games and Economic Behavior, 68(1), 95–107.

    Article  Google Scholar 

  16. Falk, A., & Fischbacher, U. (2006). A theory of reciprocity. Games and Economic Behavior, 54(2), 293–315.

    Article  Google Scholar 

  17. Fehr, E., & Schmidt, K. (1999). A theory of fairness, competition, and cooperation. The Quarterly Journal of Economics, 114(3), 817–868.

    Article  Google Scholar 

  18. Fehr, E., Klein, A., & Schmidt, K. (2007). Fairness and contract design. Econometrica, 75(1), 121–154.

    Article  Google Scholar 

  19. Fehr, E., Kirchler, E., Weichbold, A., & Gächter, S. (1998). When social norms overpower competition—gift exchange in labor markets. Journal of Labor Economics, 16, 324–351.

    Article  Google Scholar 

  20. Fischbacher, U. (2007). Z-tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10, 171–178.

    Article  Google Scholar 

  21. Gneezy, U. (2005). Deception: the role of consequences. The American Economic Review, 95, 384–394.

    Article  Google Scholar 

  22. Goeree, J. K., & Yariv, L. (2011). An experimental study of collective deliberation. Econometrica, 79(3), 893–921.

    Article  Google Scholar 

  23. Goeree, J. K., & Zhang, J. (2013a). Promises and partnerships revisited. Working Paper, University of Zürich.

  24. Goeree, J. K., & Zhang, J. (2013b). Market talk. Working Paper, University of Zürich.

  25. Grosskopf, B. (2003). Reinforcement and directional learning in the ultimatum game with responder competition. Experimental Economics, 6, 141–158.

    Article  Google Scholar 

  26. Huck, S., Lünser, G. K., & Tyran, J. R. (2012). Competition fosters trust. Games and Economic Behavior, 76(1), 195–209.

    Article  Google Scholar 

  27. Kartik, N. (2009). Strategic communication with lying costs. Review of Economic Studies, 76(4), 1359–1395.

    Article  Google Scholar 

  28. Rabin, M. (1993). Incorporating fairness into game theory and economics. Journal of Economic Theory, 48(1), 179–220.

    Google Scholar 

  29. Rode, J. (2010). Truth and trust in communication: experiments on the effect of a competitive context. Games and Economic Behavior, 68, 325–338.

    Article  Google Scholar 

  30. Roth, A. E., Prasnikar, V., Okuno-Fujiwara, M., & Zamir, S. (1991). Bargaining and market behavior in Jerusalem, Ljubljana, Pittsburgh, and Tokyo: an experimental study. The American Economic Review, 81(5), 1068–1095.

    Google Scholar 

  31. Serra-Garcia, M., Damme, E. V., & Potters, J. (2011). Hiding an inconvenient truth: lies and vagueness. Games and Economic Behavior, 73, 244–261.

    Article  Google Scholar 

  32. Sutter, M. (2009). Deception through telling the truth?! Experimental evidence from individuals and teams. Economic Journal, 119, 47–60.

    Article  Google Scholar 

  33. Stigler, G. (1987). Competition. In J. Eatwell, M. Milgate, & P. Newman (Eds.), The new Palgrave. A dictionary of economics (pp. 531–535). London: The MacMillan Press Limited.

    Google Scholar 

  34. Vanberg, C. (2008). Why do people keep their promises? An experimental test of two explanations. Econometrica, 76(6), 1467–1480.

    Article  Google Scholar 

  35. Zhang, J. (2013). Communication in asymmetric group competition over public goods. Working Paper, University of Zürich.

Download references

Author information



Corresponding author

Correspondence to Jingjing Zhang.

Additional information

We gratefully acknowledge financial support from the Swiss National Science Foundation (SNSF 135135) and the European Research Council (ERC Advanced Investigator Grant, ESEI-249433). We thank Gary Charness and Martin Dufwenberg for sharing their data and valuable insights and Kremena Valkanova for excellent research assistance. We benefitted from helpful comments made by the special editor, Tore Ellingsen, two anonymous referees, as well as seminar participants at the Economic Science Association meetings in Tucson (November 2010) and the “Communication in Experimental Games” workshop in Zürich (June 2011).

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

(PDF 595 kB)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Goeree, J.K., Zhang, J. Communication & competition. Exp Econ 17, 421–438 (2014).

Download citation


  • Cheap talk
  • Adverse selection
  • Competition
  • Guilt aversion
  • Lie aversion
  • Inequality aversion
  • Reciprocity

JEL Classification

  • C92