Skip to main content
v2026.11,986 entries · CC-BY 4.0

Public Goods Game: Payoff Formula, MPCR and a Worked Example

The public goods game (voluntary contribution mechanism): payoff formula, MPCR and MGR, a worked four-player example, reporting and code.

Written and maintained by CASRAI Editorial Board

Published

On this page

The public goods game is a laboratory experiment in which each member of a group privately chooses how much of an endowment to contribute to a shared pot; the pot is multiplied and split equally among all members, whether or not they contributed. Because each token kept is worth more to the individual than the share of a token contributed, the payoff-maximising choice for a self-interested player is to contribute nothing, even though everyone is better off if all contribute. Experimenters use it, usually under the name voluntary contribution mechanism (VCM), to measure cooperation, free riding and the effect of rules such as punishment or group size.

Definition: what the public goods game (voluntary contribution mechanism) is

In the standard linear design, n players each receive an endowment E. Each player i chooses a contribution ci between 0 and E and keeps the rest. All contributions go into a group account, and every player receives the same return from it. Free riding is the term for collecting that return without contributing. The game is an n-player generalisation of the prisoner’s dilemma studied in game theory: individually rational choices produce a collectively worse outcome than cooperation would.

Experimental economists use the label “voluntary contribution mechanism” for the same set-up, and describe it as a main topic of experimental economics (Xu, Cadsby, Fan and Song, 2011). Its place in the wider field is covered in what behavioural economics is; the field’s publication venues are listed in the Journal of Experimental Economics guide.

Public goods game payoff formula and symbols

The payoff to player i in the linear public goods game is:

πi = E − ci + m × Σj cj

  • πi: payoff of player i at the end of the round, in tokens.
  • E: the endowment each player receives at the start of the round.
  • ci: player i’s contribution, with 0 ≤ ci ≤ E.
  • Σj cj: the total contributed by all n players, including player i.
  • m: the marginal per capita return (MPCR), the payoff each player receives for each token anyone contributes.

The companion quantity is the marginal group return, MGR = n × m, which is what one contributed token is worth to the group as a whole. In the design reported by Xu et al. (2011), a four-person group with MPCR 0.4 has an MGR of 0.4 × 4 = 1.6, so one token contributed yields 1.6 tokens divided equally. Two conditions define the dilemma, and both follow from the formula:

  • Free riding pays when m less than 1. Raising ci by one token changes player i’s payoff by −1 + m, which is negative, so contributing nothing is the dominant choice and complete free riding is the unique Nash equilibrium of the stage game.
  • Cooperation is efficient when n × m greater than 1. A token contributed raises group payoff by n × m − 1, so full contribution beats full free riding.

Fehr and Gächter (2000, 2002) report that complete free riding remains the unique stage-game Nash equilibrium both with and without a punishment option, as summarised by Xu et al. (2011).

Worked example: four players, endowment 20, MPCR 0.4

Take n = 4, E = 20 tokens and m = 0.4, so n × m = 1.6. The arithmetic for each scenario is πi = 20 − ci + 0.4 × (group total).

Scenario Contributions (P1, P2, P3, P4) Group total Payoffs (P1, P2, P3, P4)
Nobody contributes 0, 0, 0, 0 0 20, 20, 20, 20
Everybody contributes everything 20, 20, 20, 20 80 32, 32, 32, 32
P1 free rides on three contributors 0, 20, 20, 20 60 44, 24, 24, 24
P1 contributes alone 20, 0, 0, 0 20 8, 28, 28, 28
Everybody contributes half 10, 10, 10, 10 40 26, 26, 26, 26

Row three: P1 receives 20 − 0 + 0.4 × 60 = 44, while each contributor receives 20 − 20 + 0.4 × 60 = 24. Row four: P1 receives 20 − 20 + 0.4 × 20 = 8, below the 20 P1 would keep by contributing nothing. A single token shows the incentive directly: contributing it costs P1 one token and returns 0.4 tokens, a net change of −0.6, while the group gains 1.6 − 1 = 0.6 tokens in total. The best outcome for the group (32 each) is not an equilibrium, and the equilibrium (20 each) is worse for everyone.

Interpreting results: contribution, free riding and efficiency

Studies report average contribution as a share of the endowment, usually per round, and compare it with two benchmarks computed from the formula.

Observed pattern How to read it
Average contribution near 0% of E Behaviour matches the stage-game Nash prediction of complete free riding when m less than 1.
Average contribution near 100% of E Behaviour matches the group-efficient outcome when n × m greater than 1; each player earns n × m × E (32 in the worked example).
Contribution between the two Neither benchmark describes the group. Chaudhuri (2011) surveys the mechanisms that let conditional cooperators sustain high contributions: costly monetary punishment, expressions of disapproval, advice and assortative matching.
Contribution rises when punishment is added Fehr and Gächter (2000) show that giving players information on peers’ contributions plus costly punishment sustains high contributions in groups of four, under both partner and stranger designs.

Group size and the MPCR confound

A frequent mistake is to compare a group of 4 with a group of 40 at the same MPCR. Holding m fixed while n rises raises MGR (n × m), so the larger group faces a stronger incentive to cooperate as well as a larger group. Isaac, Walker and Williams (1994) studied groups of 4 to 100 with MPCRs from 0.04 to 0.75: at MPCR 0.3, groups of 40 and 100 provided the public good more efficiently than groups of 4 and 10, while at MPCR 0.75 group size had no significant effect (as reported by Xu et al., 2011). Xu et al. therefore compare 40-person groups at MPCR 0.4 (MGR 16) and MPCR 0.04 (MGR 1.6, matching the four-person design) and report that punishment keeps much but not all of its effectiveness when the MPCR shrinks with group size.

Assumptions and common mistakes

  • Linear payoffs. The formula above assumes the group return is proportional to total contributions. Other designs (threshold public goods, non-linear returns) need their own equilibrium analysis.
  • Self-interested players assumed only for the benchmark. The Nash prediction is a reference point, not a claim that participants are payoff-maximisers; the reported data are about how far behaviour departs from it.
  • Partner versus stranger matching. Fehr and Gächter’s results are reported under both; state which protocol you used because repeated play with the same group is not the same design as re-matching each round.
  • Confounding group size with MGR. See above: vary n and m deliberately, and report n × m.
  • Punishment stage payoffs. Report whether punishment is costly to the punisher and whether income can fall below zero; in Xu et al.’s design, income at the contribution stage cannot be pushed below zero by punishment.
  • Round-by-round data are not independent. Rounds within a group are correlated, so analyse at the group level or use methods that account for clustering. When you report results, pair significance tests with an effect size.

How to report a public goods experiment (APA 7th edition style)

State the design parameters first, then the outcome. A template (illustrative wording, replace the bracketed values with your own results):

Participants played a [10]-round linear public goods game in groups of [4] under a [partner] matching protocol. In each round, participants received [20] tokens and chose a contribution; each token contributed returned [0.4] tokens to every group member (MPCR = [0.4], MGR = [1.6]). Mean contribution was [M = …, SD = …] tokens in Round 1 and [M = …, SD = …] in Round 10.

Cite the source of the design in the reference list in APA format, for example: Fehr, E., & Gächter, S. (2000). Cooperation and punishment in public goods experiments. American Economic Review, 90(4), 980–994. https://doi.org/10.1257/aer.90.4.980. Planning the number of groups is a sample-size question at the group level; see sample size determination formulas, and categories of research design for where experiments sit among designs.

How to calculate public goods game payoffs in Excel, R or Python

Excel

Put contributions in B2:B5. In C2, enter the payoff for the first player and fill down. The endowment is 20 and the MPCR is 0.4:

=20 - B2 + 0.4 * SUM($B$2:$B$5)

R

payoff = function(c, E = 20, m = 0.4) E - c + m * sum(c)
payoff(c(0, 20, 20, 20))   # 44 24 24 24

Python

def payoff(c, E=20, m=0.4):
    total = sum(c)
    return [E - x + m * total for x in c]

print(payoff([0, 20, 20, 20]))   # [44.0, 24.0, 24.0, 24.0]

Sources

Data: Public goods game group size, MPCR and marginal group return in the Xu et al. (2011) designs

Group size, marginal per capita return (MPCR) and the derived marginal group return (MGR = group size x MPCR) for the three voluntary contribution mechanism designs reported in a University of Guelph working paper on group size and punishment.

Period covered: 2011.

Public goods game group size, MPCR and marginal group return in the Xu et al. (2011) designs
DesignGroup size (players)MPCR (tokens per token contributed)MGR (tokens per token contributed)
4-person group, MPCR 0.440.41.6
40-person group, MPCR 0.4400.416
40-person group, MPCR 0.04400.041.6

Sources

Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.

Follow CASRAI

Research-administration guidance, standards updates and independent tool reviews.

Ask CASRAI · free to try

Ask about Public Goods Game: Payoff Formula, MPCR and a Worked Example

2 free questions a day, no account.

An AI assistant for research: funder rules, compliance, methods and publishing. It links the sources behind its answers and labels answers that come from a web search.

Answers draw on CASRAI's guides and dictionary plus the federal and funder documents we index: Federal Register, Grants.gov, Regulations.gov and UKRI.

Works on this site and inside Claude, Cursor and the AI tools you already use.

Everything CASRAI publishes — this page, the dictionary, the guides and the news — stays open to read, with no account and no card.

Referenced across the research world

University of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logoUniversity of Cambridge logoColumbia University logoCrossref logoUniversity of Edinburgh logoHarvard University logoUniversity of Oxford logoPrinceton University logoStanford School of Medicine logoUniversity College London logoORCID logo
  • University of Cambridge logo
  • Columbia University logo
  • Crossref logo
  • University of Edinburgh logo
  • Harvard University logo
  • University of Oxford logo
  • Princeton University logo
  • Stanford School of Medicine logo
  • University College London logo
  • ORCID logo

View CASRAI adoption →