Changing partners in a cheap talk game: Experimental evidence

• DOI: http://doi.org/10.1111/ijet.12125
• Author: Olivier Bonroy,Alexis Garapin,Daniel Llerena
• Published date: 01 June 2017
• Date: 01 June 2017

doi: 10.1111/ijet.12125

Changing partners in a cheap talk game:

Experimental evidence

Olivier Bonroy,∗Alexis Garapin†and Daniel Llerena‡

This paper considers the effects that the opportunity to change partners has on communication.

Our experiment concerns a standard cheaptalk game in which a player observes a private forecast

before disclosing it (truthfully or untruthfully) in a message that he sends to his partner. Two

treatments are applied: in one, each team remains unchanged until the experiment ends; and

in the other, players can change their partner.We find that the opportunity to change partners

positively affects communication in the relationship. Interestingly, this effect is explained by

higher levels of trust in the messages and not by more truthful disclosure of private information.

Key wor ds cheap talk, information disclosure, monitoring, partner, experimental economics

JEL classification C72, C90, D82

Accepted 14 April 2016

1 Introduction

In business relationships, an agent (a manager,a retailer, or a plaintiff) may have private information

on an opportunity that he/she may pass along to his/her business partner (an investor, a supplier, or

a defendant) via cheap talk (Ren et al. 2010). When the business partners’ interests are sufficiently

conflicting, the owner of the private information may have no interest in sharing the information:

cheap talk is meaningless. Long-term relationships may then enhance the credibility of cheap talk

so that information sharing is trustworthy. However, even in the long run, partners in business

relationships are likely to change, and it is difficult to believe that such an eventuality can have no

influence on information sharing.

Let us take the example of a retailer and a supplier. The retailer is assumed to have better

knowledge of the final demand than his/her supplier because of his/her supposed proximity to the

final market. The supplier must often rely on the retailer’sforecasts to allocate production capacities

in order to deliver goods. As the forecasts only provide information about the downstream firm’s

intentions regarding a given future state of the world (there is no explicit contract), they can be

considered as “cheap talk” messages (Rene t al. 2010). This forecast-information asymmetry creates

∗INRA, Grenoble Alpes University,GAEL, Grenoble, France.

†Grenoble Alpes University,INRA, GAEL, Grenoble, France. Email : alexis.garapin@univ-grenoble-alpes.fr

‡Grenoble Alpes University,INRA, GAEL, Grenoble, France.

Wewish to thank Jean-Loup Dupuis for technical assistance with the computer laboratory and Liz Libbrecht for editorial

assistance. Wethank an anonymous referee for useful suggestions that improved the paper.We also acknowledge P.Burlat,

X. Boucher, N. Taratynava (Ecole des Mines de Saint Etienne), the participants in various conferences (including ESA

2011) and seminars, for discussion and comments.This research received financial support from the R´

egion Rhˆ

one-Alpes,

program “COPILOTES”.

International Journal of Economic Theory 13 (2017) 197–216 © IAET 197

International Journal of Economic Theory

Changing partners in a cheap talk game Olivier Bonroy et al.

an incentive problem: the retailer may then use his/her private information advantage to inflate the

level of the final demand in order to ensure an abundant supply, that is, to deceive.1However, in a

long-term relationship, the retailer may have an interest in sharing the information truthfully: this

is the reputation effect. Possible current gains from opportunistic behavior are then wiped out by

future losses in payoff from damaged reputation (see, for example, Kim 1996). Assume now that

both retailer and supplier may break up the business relationship by changing partners, as can be

seen in the automotive or agribusiness industry. How can an opportunity to change partners impact

the forecast-information sharing?

Theoretically,infinite repetition of the relationship is a necessary condition so that the reputation

effect holds (see, for example, Tirole 1988; Kim 1996). In this way, if the opportunity to change

partners is considered by agents as a possible end of the relationship, then it modifies the infinite

character of the repetition of the relationship such that in each period the retailer has no interest

in truthfully disclosing his/her private information. The retailer is not concerned about his/her

reputation and the cheap talk is meaningless.

In this paper, we consider an experiment showing how private information sharing behaviors

are affected by an opportunity to change partners in such cheap talk settings. We consider a cheap

talk game that we repeat in a context of imperfect monitoring and change of partners. Some of the

literature in experimental economics has introduced the possibility of changing game partners in

strategic settings, showing that it has significant effects on cooperation. But this possibility has not

yet been examined in cheap talk games. Davis and Holt (1994) conducted an experiment in which

subjects were placed in three-person, choice-of-partner games. The subject who got to choose the

other two partners to play with was able to administer a salient punishment, switching partners, in a

multistage game. Their results show that if the “chooser”decides to switch partners after experiencing

defection, the subsequent level of cooperative play bythe players selected is higher than in the control

treatment (without the switching option). The longer the multistage game, the higher the level of

cooperation will be after the switching option is taken. Hauk and Nagel (2001) and Coricelli et al.

(2004) study the effect of a unilateral and mutual choice of partners, the former in finitely two-person

repeated prisoner’s dilemma games and the latter in finitely two-person public good games. In both

games, results demonstrate significantly higher cooperation in unidirectional partner selection than

in bidirectional partner selection and random rematching. Barclay and Willer (2007) also studied

experimentally different variants of selecting another partner in a three-player prisoner’s dilemma

game setting. Their results demonstrate that the levels of cooperation in the prisoner’s dilemma

game are highest when players know that they are likely to be selected bythe third player to carry on

playing the game.

In every period in our experimental game, the sender observes private information on a business

opportunity before disclosing it (truthfully or untruthfully) in a message that he sends to the receiver.

The receiver determines an investment level by choosing to believe the message or not, before the

business opportunity is realized and revealed to both players. Thus, the receiver’s decision is payoff-

relevant to both players. Note that the veracity of the message will be revealed or not, depending

on the realization of the business opportunity. This imperfect monitoring setting is motivated by

its ability to stress the sender’s incentive for deception, and thereby to make it easier to quantify

1In real-world business, examples of this behavior havebeen reported in the press. For instance, in March 2001, Solectron,

a leading electronics supplier (acquired in 2007 by Flextronics), had to write off $4.7 billion in excess inventory due

to inflated customer forecasts (Business Week, March19, 2001). In November 2004, Boeing suffered a $2.7 billion loss

because American Airlines deferred the purchase of 54 of 56 aircraft initially scheduled for delivery between 2006 and

2010 (WallStreet Journal, November 22, 2004).

198 International Journal of Economic Theory 13 (2017) 197–216 © IAET