Probabilistic continuation in the public goods game

How to cite

Kiss, Hubert Janos; Pinter, Agnes; Kovarik, Jaromir. 2018. "Probabilistic continuation in the public goods game." AEA RCT Registry. AEARCTR-0003497. https://doi.org/10.1257/rct.3497-1.0

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

Hubert Janos Kiss* , Agnes Pinter , Jaromir Kovarik

* Primary investigator

Registry ID
AEARCTR-0003497
DOI
https://doi.org/10.1257/rct.3497-1.0
First registered
Last updated
Status
In development
Trial dates
2018-11-10 – 2019-12-10
Countries
Private
JEL codes
C91
Registry record
http://www.socialscienceregistry.org/trials/3497

Abstract

In the classic public goods game (PGG) subjects usually play a given number of repetitions, say 10 rounds. The number of rounds is independent of the subjects' contribution. We argue that in many cases the provision of a public good depends on past contributions, so we assume that the next round is reached only in a probabilistic manner: the more the subjects contribute, the more likely it is that there will be a next round.
We study if such a probabilistic setting changes subjects' behavior relative to the classic setup. We use an experimental design with parameters that theoretically should lead to zero contribution at the beginning of the game that in turn would mean that the game ends in round 1. However, we expect that many groups will play several rounds and there will be a high level of overall contribution in many groups.

Intervention

In the baseline treatment, we replicate the classic four-person public goods game (PGG) with MPCR=0.4 and 10 rounds. In each round, each player starts with an endowment of 25 ECUs that she can split between a private and a public account. We expect to see the well-known pattern of 40-60% contributions in the first round and then a steady decline.

In the treatment that we call dependent, the probability of reaching the next round depends on the total contributions. Note that the sum of the initial endowments is 100 ECUs and the total contribution as share of these 100 ECUs determines the probability of going to the next round. Hence, if total contribution is 100 / 63 / 0, then the game proceeds with 100% / 63% / 0% probability.

Experimental design

The experiment has two parts: a classic and a modified public goods game (PGG). We will have four sessions with 60 subjects in each of them. In all sessions, both of the PGGs will played, but in different order. In half of the sessions, participants start with the classic PGG, followed by the modified one, while in the other half of the sessions it will be the other way around. Importantly, at the beginning of the sessions we do not let them know that there will be two games, the second one is a surprise. In this respect, we are followingthe design by Fehr, E., & Gachter, S. (2000). Cooperation and punishment in public goods experiments. American Economic Review, 90(4), 980-994.

In sessions 1 and 2, subjects first play the classic PGG and then the modified one. After entering the labe and being seated, subjects receive the first part of the instructions that describes the first part of the experiment. They play the classic PGG with 10 rounds. aAfter finsihing they are told that they have the opportunity to play another similar game and earn more money. They are given the instructions for the second part that describes the modified PGG. They play this part of the experiment and after finsihing they fill in the questionnaire. After all, they receive their earnings.
In sessions 3 and 4, the experiment starts with the modified PGG, followed by the classic one. The course of the experiment is the same as before.
The exchange rate is 50 ECUs=1 euro and we pay a show-up fee of 5 euros.

While in the classic PGG subjects play 10 rounds with certainty, in the modified PGG the probability of reaching the next round is determined by the total contribution. It may be the case that a group "dies" early due to low contribution in the early rounds. Since this modified PGG is more complicated than the clasicc PGG, we want to give them more possibilities to play, so subjects may play the modified PGG five times. Note that possibly the game ends before round 10.

The final payment to the subjects consists of the following:
- earnings in the classic PGG
- earnings in one of the randomly chosen repetitions in the modified PGG
- show-up fee.

Design details

Randomization method
Each subject plays both the classic and the modified PGG. It will be random in which order they will play them and the randomization is carried out by the lab (LINEEX, University of Valencia) that runs the experiment for us.
Randomization unit
Experimental session.
Clusters
2 sessions with 60-60 subjects in which the order of the experiment is classic PGG fllowed by the modified one. 2 sessions with 60-60 subjects in which the order of the experiment is modified PGG fllowed by the classic one.
Observations
We will have in total 240 subjects. Each will play the classic PGG that implies 10 decisions per subjects, that is 2400 decisions. All of them will play also the modified PGG five times. The minimum number of decisions in these five repetitions are 5 and the maximum 50, so we will have in this part between 1200 and 12000 observations.
Arms
2 sessions with 60-60 subjects in which the order of the experiment is classic PGG fllowed by the modified one. 2 sessions with 60-60 subjects in which the order of the experiment is modified PGG fllowed by the classic one. In total, we will have 240 subjects.
Primary outcomes
The key variables are the individual and the total contributions over the rounds. Moreover, since continuation in the dependent treatment depends on total contribution, we are also interested in how many rounds are played.
Secondary outcomes
Since playing the public goods game (PGG) with the probabilistic continuation may be difficult at the first time for players (as it is obviously more complicated than the classsic PGG), we let them play the game five times. We are interested in seeing how they learn to play the game (if there is learning going on at all).

Cited by

No citation data available for this registration yet.

Keywords