Choosing the agent's group identity in a trust game with delegated decision making

• Published date: 01 February 2020
• Date: 01 February 2020
• Author: Nicky Nicholls,Alexander Zimper
• DOI: http://doi.org/10.1111/jpet.12405

J Public Econ Theory. 2020;22:220–244.wileyonlinelibrary.com/journal/jpet220

|

© 2019 Wiley Periodicals, Inc.

Received: 22 February 2019

|

Accepted: 23 September 2019

DOI: 10.1111/jpet.12405

ORIGINAL ARTICLE

Choosing the agent’s group identity in a trust

game with delegated decision making

Alexander Zimper

|

Nicky Nicholls

Department of Economics, University of

Pretoria, Pretoria, South Africa

Correspondence

Alexander Zimper, Department of

Economics, University of Pretoria, Private

Bag X20, Hatfield, Pretoria 0028, South

Africa.

Email: alexander.zimper@up.ac.za

Abstract

Members of a given social group often favor members of

their own group identity over people with different

group identities. We construct a trust game in which the

principal delegates the decision about an investment

into a receiver to an agent who either favors the

principal’s or the receiver’s group identity. When

choosing the agent’s group identity the principal faces

a trade‐off between a loyal agent and an agent who

might increase the receiver’s willingness to cooperate.

We solve for the principal’s decision in a subgame‐

perfect nash equilibrium for the two scenarios of a risk‐

neutral and risk‐averse agent, respectively.

KEYWORDS

fairness, identity, in‐group bias, trust game

JEL CLASSIFICATION

C72; D82

1

|

INTRODUCTION

Starting with Tajfel (1970) there exists extensive empirical literature at the intersection between

psychology, sociology, and behavioral economics which demonstrates that considerations about

group identity play an important role in individual preferences. A persistent finding in this

literature is manifestations of an “in‐group bias”according to which an individual of a given

social group tends to favor—or to trust—individuals of her own group identity over individuals

that belong to a different social group (see, e.g., Balliet & Van Lange, 2013; Chen & Li, 2009;

Currarini & Mengel, 2016; Kuwabara et al., 2007; Morton, Ou, & Qin, 2019; Sherif, 2010; Tajfel

& Turner, 1986). Experiments that establish the existence of an “in‐group bias”in strategic

situations typically take the form of dictator games—where one individual can allocate money

to another individual—or of trust games—where one individual can invest money into another

individual with the prospect of receiving some money in return.

1

Applied to a situation in which a principal has to choose between employing either an agent

of her own or of a different group identity, the existence of an “in‐group bias”seems to suggest

that the principal will always favor the agent of her own over the agent of a different identity.

However, this seemingly straightforward relationship between identity considerations and

employment decisions does not capture the more complex—but relevant—strategic situations

in which an agent with an in‐group bias has to interact on behalf of the principal with a receiver

who cares about the agent’s group identity. Examples of such situations with delegated

decision‐making that involve a receiver who is of different group identity than the principal are

aplenty. The original idea to our model goes back to a discussion about the historical question of

why the British East Indian Company (principal) might have chosen gomasthas (agents with

local identity) rather than her own British employees to deal on her behalf with the local

population (receiver). But there exist more mundane everyday situations where group identity

considerations concerning the agent and the receiver are relevant to the principal and which,

therefore, fall broadly under our modeling framework.

2

This paper develops a theoretical model about the possible impact of group identity

considerations on the principal’s employment decision in a strategic environment that involves

the three players “principal,”“agent,”and “receiver,”respectively.

3

The “principal”of given

group identity has to bankroll an investment into a “receiver”of different group identity. The

principal will only receive a positive return on this investment if the receiver cooperates

towards the principal’s objectives. We further assume that the principal has to delegate the

decision about this investment to an “agent”whereby the principal cannot impose any

enforceable contract about the agent’s decision. As her only decision, the principal can either

choose an agent of her own or of the receiver’s group identity. The choice of the agent’s identity

matters to the principal for two reasons. First, the agent has a strong in‐group bias: whereas the

agent with the principal’s identity favors the principal, the agent with the receiver’s identity

favors the receiver. Second, by choosing the agent’s identity, the principal might be able to

influence the likelihood that the receiver will be of a cooperative type. In a nutshell, our model

thus describes a situation in which the principal faces the trade‐off between choosing an agent

who is loyal to her versus an agent who is not loyal but might positively influence the receiver’s

willingness to cooperate.

Intuition suggests that the principal should choose the “disloyal”agent with the receiver’s

identity only if this agent sufficiently increases the likelihood of a cooperative receiver. To make

this intuition precise—and to go beyond it—we construct a trust game with delegated decision

making for which we characterize the principal’s choice as the outcome of a subgame‐perfect

nash equilibrium (SPNE) in dependence on the model’s payoff parameters. Recall that the

original trust game is a two‐stage game in which the principal decides about the amount

a

of

1

For early discussions of trust games see Berg, Dickhaut, and McCabe (1995) and references therein. For more recent studies see, for example, Glaeser, Laibson,

Scheinkman, and Soutter (2000), Fershtman and Gneezy (2001), Ashraf, Bohnet, and Piankov (2006), Burns (2006), Güth, Levati, and Ploner (2008), Bornhorst,

Ichino, Kirchkamp, Schlag, and Winter (2010), Chowdhury, Jeon, and Ramalingam (2016), Weisel and Zultan (2016), Daskalova (2018), Hamann and Nicholls

(2018). Whereas Johnson and Mislin (2011) provide a meta‐analysis for trust game experiments, Buchan et al. (2002) discuss the empirical evidence of an “in‐

group bias”for a great variety of strategic set‐ups, including dictator, as well as trust games.

2

To quote an associate editor: “The white CEO who has staff that are mainly black—should he hire a white or black store manager? The charity trying to raise

money in the US for disaster relief victims in Africa—who to hire as fundraisers? The microfinance bank that has rich managers but wants to appeal to lower

income families—should she hire a rich or poor local representative? The white store manager trying to sell products to black or Asian customers—who to hire

as store reps?”

3

To distinguish between these players, we refer throughout the paper to the “principal”as she, to the “receiver”as he, and to the “agent”as it.

ZIMPER AND NICHOLLS

|

221