Cooperation in a risky world

• DOI: http://doi.org/10.1111/jpet.12366
• Published date: 01 April 2020
• Author: Vincent Théroude,Adam Zylbersztejn
• Date: 01 April 2020

Received: 14 September 2017

|

Revised: 20 November 2018

|

Accepted: 3 February 2019

DOI: 10.1111/jpet.12366

ORIGINAL ARTICLE

Cooperation in a risky world

Vincent Théroude

|

Adam Zylbersztejn

Univ Lyon, Université Lyon 2, GATE UMR

5824, F‐69130 Ecully, France

Correspondence

GATE (Groupe d'Analyse et de Théorie

Economique), CNRS ‐University of Lyon 93,

Chemin des Mouilles, 69130 Ecully, France.

Email: zylbersztejn@gate.cnrs.fr

Funding information

University of Lyon 2, Grant/Award Number:

AIP 2016

Abstract

We offer a novel investigation of the effect of environmental

risk on cooperation in the Voluntary Contribution Mechanism.

Our baseline is the standard setting, in which the personal

return from the public good is deterministic, homogeneous, and

publicly known. Our experimental treatments alter this classic

design by making the marginal per capita return from the public

good probabilistic. In the homogeneous risk (HomR) treatment,

the random draw is made for the whole group, whereas in the

heterogeneous risk (HetR) treatment, this happens independently

for each group member. Our hypothesis is that different

environmental risks may differently affect the ex post payoff

inequalities, so that other‐regarding preferences (inequality

aversion) may generate higher contributions in HomR than in

HetR. Our main result is that the environmental risk does not

affect the patterns of cooperation either in the one‐shot or in

the finitely repeated version of the game. This suggests that the

standard experimental methodology provides a robust and

conservative measure of human cooperation.

1

|

INTRODUCTION

Many economic interactions create tension between individual and collective interests. For this reason, human

cooperation has become one of the central topics in behavioral economics. Since the pioneering work of Isaac,

McCue, and Plott (1985), numerous economists have studied cooperation using the Voluntary Contribution

Mechanism (VCM).

1

Over time, VCM has become a widely accepted experimental testbed for studying various

environmental and institutional aspects of cooperation (Chaudhuri, 2011; Ledyard, 1995).

Strikingly, what has once evolved to become the “gold standard”in experimental economics—that is, a version

of VCM based on deterministic, homogeneous, and publicly known personal returns from the public good—seems

far off as compared not only to the seminal work by Isaac et al. (1985) but also to many real‐world situations. Isaac

J Public Econ Theory. 2020;22:388–407.wileyonlinelibrary.com/journal/jpet388

|

© 2019 Wiley Periodicals, Inc.

1

According to another early study by Isaac, Walker, and Thomas (1984), this seminal work dates back to 1980. Ledyard (1995) lists the study by Isaac

et al. (1985) as the first economic experimental on public goods, along with Kim and Walker (1984), who conducted their experiments 2 months later.

et al. (1985) use heterogeneous returns from the public good, which are entirely private knowledge, a design that

certainly adds realism to the experimental game.

2

For instance, the benefits that the individuals derive from

establishing public facilities for health care, education, or social security are likely to be heterogeneous, private

knowledge, and furthermore subject to randomness and variation over time.

Yet, relatively little is still known about how the patterns of cooperation change once the decision‐making

environment shifts away from the “gold standard”paradigm. Addressing this issue seems important in terms of both

the internal and the external validity of laboratory experiments. First, the data from alternative settings can be

informative about the robustness and the limitations of the large body of experimental findings based on the

standard VCM setup. Second, as recently documented in a carefully crafted lab‐in‐the‐field experiment by Stoop,

Noussair, and van Soest (2012), the usual patterns of cooperation observed in the standard VCM may not prevail in

analogous (yet involving more complexity and uncertainty) real‐world environments.

3

Explaining such discrepancies

and rendering the findings from lab experiments applicable to naturally occurring settings calls for greater care for

the ecological validity of laboratory experiments. Our experiment is a step in that direction.

Herein, we tweak the standard VCM by incorporating two kinds of environmental risk, which may occur either at

the individual or at the group level. In both cases, the personal return from the public good is not deterministic, but

probabilistic—either higher or lower than the riskless one. Both outcomes are equiprobable, become known ex post

(i.e., only after the contribution decisions have been made), and the lottery is mean‐preserving with respect to the

standard VCM scenario. In the homogeneous risk (HomR) treatment, the random draw is made in each round for the

whole group. In the heterogeneous risk (HetR) treatment, in turn, this happens in each round independently for each

group member. Thus, our experimental paradigm provides a simple way to capture the distinction between local

and global environmental risks that are important in various real‐world public good settings.

4

Our experiment

complements the previous work by Lévy‐Garboua, Montmarquette, Vaksmann, and Villeval (2017), who study

cooperation in building a common insurance pool. In their design, agents face the risk of losing their resources and

ex ante may voluntarily fund a group insurance scheme to help the needy members recover their losses. Herein, the

risk is related to the ex post capacity to benefit from a common resource.

Our experiment consists of two incentivized VCM‐based tasks. In the first part, we elicit players’conditional

contributions to the public good, their unconditional contribution in a one‐shot interaction, as well as their beliefs

about other group members’contributions. In the second part, the participants play a finitely repeated VCM game

under partner matching. We implement environmental risk treatments in a between‐subject manner. Given the

proper randomization of social and risk preferences across experimental conditions, our investigation provides

causal evidence on the effect of environmental risk on different layers of human cooperation.

Our theory builds on the insight from M. Krawczyk and Le Lec (2010) that social interactions in risky

environments are affected by both distributive fairness (i.e., ex post outcomes) and procedural fairness (i.e., ex ante

opportunities). Here, our intuition is that although different kinds of environmental risk preserve the expected

2

Of course, using the standard design has major advantages. First, it reduces the degree of unwarranted uncertainty in the decision‐making environment,

thus improving the experimenter’s control over subjects’decisions. Second, it makes the outcomes of different experiments easier to compare and

replicate, thus fostering the accumulation of empirical knowledge.

3

They have conducted an experiment among recreational fishermen at a private pond. There are two variations of a VCM game: a framed field game and a

standard lab game. In both tasks, fishermen decide how many fish they want to catch. Abstaining from catching a fish (and thus reducing one’s own

welfare) generates positive externalities (which exceed that individual welfare loss) for other fishermen. They report that the usual patterns of

cooperation observed in the lab game fade away in the analogous field game. They also test (and refute) a number of possible explanations behind this

difference. However, one factor, which they do not test, is related to environmental uncertainty—in the lab VCM game, participants simply choose the

number of fish, whereas in the field, VCM game played around a fishing pond the final catch can never be certain.

4

Going back to our previous example, one’s utility from public health services may depend on idiosyncratic health hazards (such as noninfectious diseases

related to one’s genes, profession, or lifestyle), but also on the population‐wide hazards (such as large outbreaks of infectious diseases). In a similar vein,

one’s benefit from public education facilities may depend on one’s current educational needs or aspirations, but also on global labor market fluctuations

that may render education as a whole (or at least its certain types) either beneficial or redundant. Finally, the benefits a worker derives from the social

security system may depend on individual risks (such as an accident at work), but also on collective risks (such as industrial restructuring resulting in large‐

scale layoffs).

THÉROUDE AND ZYLBERSZTEJN

|

389