Discrimination via Exclusion: An Experiment on Group Identity and Club Goods

• Author: MIGUEL A. FONSECA,SURAJEET CHAKRAVARTY
• Date: 01 February 2017
• Published date: 01 February 2017
• DOI: http://doi.org/10.1111/jpet.12182

DISCRIMINATION VIA EXCLUSION:ANEXPERIMENT ON

GROUP IDENTITY AND CLUB GOODS

SURAJEET CHAKRAVARTY

University of Exeter

MIGUEL A. FONSECA

University of Exeter

Abstract

We study using laboratory experiments the impact on cooperation of al-

lowing individuals to invest in group-specific, excludable public goods.

We find that allowing different social groups to voluntarily contribute

to such goods increases total contributions. However, a significant pro-

portion of that contribution goes toward the group-specific club good

rather than the public good, even when the latter has higher finan-

cial returns to cooperation. We find significant evidence of in-group

biases, which are manifested by positive in-group reciprocity. That is,

club goods allow subjects to display their preferences for interaction

with their in-group members, as well as positive in-group reciprocity.

1. Introduction

Individuals often form groups designed to produce club goods. These goods can only be

accessed by group members to the detriment of others. Many groups that provide club

goods are institutional expressions of various forms of group identities. Membership of

those groups is often drawn along ethnic, gender, or religious lines. Examples include

some private member clubs, whose externalities are the business networks membership

provides, as well as faith schools.1

One way to understand the existence of club goods is through a social identity the-

ory framework. This theory argues that membership of social groups shapes individ-

ual preferences. This may be because group membership instills particular norms of

1For a discussion of the role of private member clubs and their increasing relevance in society, see The

Economist (2012). Faith schools are learning institutions that are typically operated and/or financed by

members of a particular faith and community for the children of that same faith or community. There

are 20,000 maintained or government-supported schools in England, of which 7000 are faith schools

(Department of Education, UK, 2010).

Surajeet Chakravarty, University of Exeter, Exeter, Devon EX4, UK (s.chakravarty@exeter.ac.uk).

Miguel A. Fonseca, University of Exeter, Exeter, Devon EX4, UK (m.a.fonseca@exeter.ac.uk).

We are very grateful to Tim Miller for his outstanding help in programming the software and running

the sessions. Financial support from ESRC grant RES-000-22-394 is gratefully acknowledged. Comments

from participants at the 2013 ESA World Meetings are gratefully acknowledged.

Received September 29, 2014; Accepted October 8, 2015.

C2016 Wiley Periodicals, Inc.

Journal of Public Economic Theory, 19 (1), 2017, pp. 244–263.

244

Discrimination via Exclusion 245

behavior (Akerlof and Kranton 2000), or because it leads individuals to have a concern

for the status of the group to which they belong (Shayo 2009). Alternatively, individu-

als may have greater concerns for the welfare of members of their group than that of

outsiders (Tajfel et al. 1971; Chen and Li 2009). As societies become more diverse, peo-

ple often rely on their communities to provide certain club goods in order to maintain

their distinct identity. As a result, we observe services such as schools, cultural, and/or

leisure activities being provided by different social groups with a view to be used by these

specific constituencies alone.

Club goods are typically close substitutes for public goods, and the former’s ex-

istence can have important implications regarding both the financing and the utility

derived from the consumption of the latter. This raises important questions: does the

ability of one or more social groups to privately provide group-specific club goods lead

to a general decline in public good provision? Are public goods perfect substitutes for

club goods or are there externalities from having multiple privately provided public

goods whose beneficiaries are different subsets of the population?

In this paper, we study the effect of allowing subjects to contribute toward a group-

specific, excludable public good in addition to a pure public good. We wish to under-

stand whether or not total contributions change as a function of the presence of the

club good, and to what extent these changes are driven by social identity preferences.

Does the membership of a social group affect the way individuals contribute to public

goods? Furthermore, do contributions to the club good work as punishments to nonco-

operative out-group members, or do they function as rewards to in-group cooperators

(or both)?

In the first part of our experiment, subjects self-select into one of two social groups

depending on their choices in a task that is completely unrelated to the main experi-

ment. In other words, we use a very weak form of group membership, in the tradition of

the minimal group paradigm (Tajfel et al. 1971). In the second part of the experiment,

subjects play a six-player cooperation game in which three subjects belong to one social

group and the other three belong to the other group. In every round of our experi-

ment, subjects receive an endowment. They can allocate any part of their endowment

in any of three ways. The first is a public account, the public good, whose returns are di-

vided equally between the six players. The second is a group-specific account, the club

good, whose returns are divided equally between the three players from their group.

Alternatively, they can keep their endowment to themselves.

We implement two main treatments: one where the marginal per capita return

(MPCR) of the public good was equal to that of the club good, and another where

the MPCR of the club good was twice that of the public good. In the first treatment,

the financial returns to cooperating are higher in the public good: although the MPCRs

are the same, there are more people contributing to the public good. In the second

treatment, the financial returns to cooperation in both club and public good are the

same, as the MPCR of the club good is normalized to the number of contributors. We

also run two controls: in one, subjects have access to a public good and a club good, but

without any group identities; in the other, we induce group identities, but subjects can

only contribute to a single pure public good.

We find that total contributions are higher when a club good is available than when

subjects can only contribute to the public good. This increase is significant when the

MPCR of the club good is normalized to group size. We find strong evidence of in-

group effects, which are manifested through positive in-group reciprocity. First, we find

positive contribution levels to the club good, even when returns to contributions to the

public good financially dominate those to the club good. Second, we find evidence that