How Do Public Goods Providers Play Public Goods Games?

• Published date: 01 May 2015
• Author: Daniel M. Butler,Thad Kousser
• DOI: http://doi.org/10.1111/lsq.12073
• Date: 01 May 2015

DANIEL M. BUTLER

Washington University in St. Louis

THAD KOUSSER

UC San Diego and Flinders University

How Do Public Goods Providers

Play Public Goods Games?

We study how policymakers play public goods games, and how their behavior

compares to the typical subjects we study, by conducting parallel experiments on college

undergraduates and American state legislators. We find that the legislators play public

goods games more cooperatively and more consistently than the undergraduates. Legis-

lators are also less responsive to treatments that involve social elements but are more

likely to respond to additional information that they receive. Further, legislators’ fixed

characteristics explain much of the variation in how legislators play the game. We dis-

cuss the implications of these findings for understanding how institutions affect the pro-

vision of public goods.

An extensive and fruitful literature has used laboratory experiments

to learn about levels of public goods provisions. These studies have pro-

duced important theoretical insights and practical solutions to signif‌icant

problems (e.g., Buchan et al. 2009; Hamman, Weber, and Woon 2011;

Milinski et al. 2006; Ostrom, Walker, and Gardner 1992, 1994). Public

goods experiments have yielded powerful insights because these

researchers have been able to directly manipulate the institutions in the

study and thus assess the causal impact of institutional variations that are

observed in the real world.

In practice, researchers typically recruit undergraduates to partici-

pate in these lab experiments; even researchers who want to make infer-

ences about the behavior of elites typically draw participants from

undergraduate subject pools.

1

This is a reasonable approach. It is more

convenient to recruit students, and they provide a baseline for how intel-

ligent and educated but otherwise ordinary human beings will act.

However, these undergraduates provide only a baseline (Kam,

Wilking, and Zechmeister 2007; Sears 1986). In practice we often

want to make inferences, based on lab experiments, about how elites

would act. Our review of recent lab experiments in the next section

LEGISLATIVE STUDIES QUARTERLY, 40, 2, May 2015 211

DOI: 10.1111 /lsq.12073

V

C2015 The Comparative Legislative Research Center of The University of Iowa

shows that many researchers, especially those studying how institu-

tions affect individuals’ interactions, explicitly draw conclusions

about elite behavior from their work. This is not surprising—the

ability to vary institutions is a great advantage of lab experimenta-

tion and scholars are interested in learning how institutions affect

elites’ behavior. Even when authors do not directly draw these con-

clusions, readers are interested in thinking about such conclusions.

For example, are the studies showing how deliberation affects levels

of cooperation among students (Ostrom, Walker, and Gardner 1992,

1994) likely to equally apply to elites?

In order to answer questions like this, we must learn whether and

how elites’ behavior differs from those of standard subject pools. Are

they more or less self‌ish? Are they more or less strategic? Are they more

or less responsive to institutional changes that affect interactions with

others? By answering such questions, we can learn when and how to

draw inferences about elites based on the play of undergraduates and

thereby make the most of lab experiments.

We provide some tentative insights into these questions by present-

ing the results of two parallel public goods experiments: one conducted

with 37 elected off‌icials from 12 American state legislatures and the

other conducted with 31 university undergraduates. The rules, experi-

mental manipulations, and incentives for the players were identical in

both games. This allows us to make a preliminary study into how these

lawmakers and college students play public goods games and how they

respond to rule changes in the game. Along some dimensions we f‌ind

important differences between the two groups.

Are these differences due to the fact that they hold elective off‌ice,

or due to their ages, career backgrounds, or any of the other ways in

which they differ from undergraduates? Without embarking on a larger

research program, we cannot isolate the impact of each part of the bundle

of attributes that differentiates students from legislators. We do not claim

to identify the causal impact of winning an election. Our purpose, rather,

is to explore how elected off‌icials play public good games under various

rules, compare their behavior with the undergraduates commonly used

in experimental samples, and see whether enough important differences

emerge to justify more research aimed at moving, in Kam, Wilking, and

Zechmeister’s (2007) words, “Beyond the ‘Narrow Data Base’.”

The experiment itself is modeled after conventional public goods

games (see Chaudhuri 2011). In the basic version of the game, players

f‌irst decide how much to withdraw from a group resource for them-

selves. The remaining amount is then doubled and shared equally among

all players in the group. Like all public goods games, and the prisoner’s

212 Daniel M. Butler and Thad Kousser