Public good with a third party

How to cite

Peeters, Ronald; Chan, Nathan; Knowles, Stephen; Wolk, Leonard. 2021. "Public good with a third party." AEA RCT Registry. AEARCTR-0007996. https://doi.org/10.1257/rct.7996-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Ronald Peeters* , Nathan Chan , Stephen Knowles , Leonard Wolk

* Primary investigator

Registry ID
AEARCTR-0007996
DOI
https://doi.org/10.1257/rct.7996-1.0
First registered
Last updated
Status
In development
Trial dates
2021-07-26 – 2022-04-25
Countries
Private
JEL codes
C72, C91, C92, D64, H41
Registry record
http://www.socialscienceregistry.org/trials/7996

Abstract

There has been extensive work in economics studying public good provision and charitable giving. A large body of work uses the voluntary contributions mechanism (VCM) game, in which small groups of players invest in a public good, the benefits of which are returned to the group members. Another large body of work uses the dictator game, in which players divide their budgets between themselves and a recipient, where the recipient may be another player or a charitable cause. These settings have different features – both from the view of canonical neoclassical theory as well as more modern behavioural economic theory – leading to different predictions for individual behaviour. Hence, fundamental questions arise about the generalizability of existing findings: How well do insights from public goods games extend to dictator games, and vice versa? And, how well do findings from experimental VCM and experimental dictator games translate into real-world charitable giving and public good provision?

In this study we construct and analyse a supergame that subsumes the VCM and charitable dictator games as special cases. The supergame structure allows us to maintain comparability between disparate settings, thus allowing for a clean examination of how prosocial behaviour differs across informative subcases. Specifically, we will vary the extent to which subjects face a dictator game, a public good game, or a mix of both, and we will test for differences in average contribution behaviour across treatments. Delving deeper, we will also examine the role of participant characteristics, including charitable giving type, gender, numeracy, etc. For each of these characteristics, we will examine whether different subtypes are more or less sensitive to treatment variations.

Experimental design

The experiment will be conducted online using oTree and participants will be recruited using Prolific.

Participants are operating within groups of size three in which they have to decide how much to contribute to a public good. They make this decision in five different situations. The overall efficiency of contributions (i.e., the total benefits generated by each contribution) is the same in all five situations. The situations differ in the share of the benefits of the public good that goes back to the three group members; the remaining share is to be donated to a charity organization (Make-a-Wish). After decisions have been made, for each group the three members are paid out according to one randomly chosen decision situation.

Before being paid out, however, participants are given three more tasks. First, a short survey eliciting their numeracy, in particular, testing their understanding of ratios, probabilities and fractions. Second, a task where we elicit their giving type (e.g., warm-glow or altruistic). Third, few questions retrieving their opinion on classifying giving behaviour to charity in the presence of “rebates” and “matches”. Both instruments are often in place to encourage charitable giving, and are equivalent, but may be understood differently.

The current idea is, for the main task, to have all participants making all five decisions; that is, the “share” being a within-subjects variation. Our plan is to collect data in two waves. After the first wave, we will briefly investigate whether the within-subjects nature produces obscure behaviour. If not, we will continue collecting the remaining observations in a within-subjects format. Otherwise, we will consider switching to a between-subjects design. All information below concerns the within-subject design, and will be updated if and when we decide to switch to a between-subjects design.

Design details

Randomization method
All randomizations are computerized.
Randomization unit
Order of decision tasks and on individual level; decision relevant for payment on group level; order of “matching” and “rebate” questions on individual level.
Clusters
1
Observations
150
Arms
150
Primary outcomes
Understand the crucial differences between public good giving (cooperation) and dictator game giving (generosity).
Secondary outcomes
Increase understanding why matching programmes are more successful in raising charitable funding than rebate programmes.

Cited by

No citation data available for this registration yet.

Keywords