Cooperative Norms and the Power of Opt-in

Giving people the freedom to choose among communities—to vote with their feet—is key to establishing and maintaining communities with cooperative norms. That’s the finding from recent experimental work by German economists Ozgur Gurek, Bernd Irlenbusch, and Bettina Rockenbach, summarized in their research paper Voting With Feet — Community Choice in Social Dilemmas.

Because the authors’ experiment is a twist on the public goods game, a brief review might be helpful. In the most basic public goods game, players receive tokens and decide, anonymously, how many tokens to contribute to a public project that generates a positive multiple of the contributions. The gains from the public project are then redistributed evenly among all players, whether they contribute or not. The socially optimal strategy is for everyone to contribute all of their tokens, but the individually optimal strategy is to contribute zero and free ride on the contributions of others.

For example, suppose I’m in a group of five players, each of which receives an initial endowment of 20 tokens, each worth one dollar. If the multiplier associated with the public project is 1.2 and we all contribute everything we have, the payoff to everyone is $24 [($20 × 5 × 1.2) / 5]. If I contribute nothing and the other subjects contribute all of their tokens, then each person receives $19.2 [($20 × 4 × 1.2) / 5]. Adding this to the $20 that I withheld from the public project, my selfishness nets me a handsome $49.2. Even if everyone else follows the same zero-contribution strategy, I’m no worse off than I was when I started.

The basic public goods game demonstrates a fundamental dilemma. We can all be better off if we cooperate, but without some enforcement mechanism, we tend to behave according to what is individually rather than socially optimal. It turns out that disclosing contribution levels and giving players an option to incur a cost to punish those who contribute little or nothing tends to increase cooperation. Even though costly punishment is irrational from a strict money-maximizing perspective, the taste for punishing uncooperative players is strong enough in some players that they incur a cost to do so. This tends to lead to higher contribution levels in subsequent rounds of the game.

Gurek, Irlenbusch, and Rockenbach reproduce the previous general finding: when people can punish bad behaviour cooperation can survive. When they can’t, cooperation collapses.

Their novel contribution is to show that letting people choose the set of rules they play under can lead to much higher levels of cooperation. The rules of the game differ in two communities. In community P, the rules state that the players contribute what they want to the public project but that all players see one another’s contributions and each player can choose to incur a cost to punish other players at the end of a round (presumably for under-contributing). In community N, players can contribute what they want, but the rules do not allow for punishment. The game persisted for thirty rounds and the players could see the payoffs in both communities after each round. In initial rounds of the game, most players chose community N. But in later rounds, nearly all subjects joined community P.

The minority of players that opted into community P in the early rounds exhibited a higher level of cooperative behavior (high contributions to the public project) and a strong willingness to punish those who did not cooperate (zero or low contributions). Players with relatively strong cooperative norms self-selected into community P because the rules there reflected their preferences, establishing and sustaining high levels of cooperation in the community. The authors find that the ability for a player to pick her community is critical. Contribution rates were much lower in control experiments that assigned people to the two communities or randomly moved people from N to P in a way that mimicked the observed movements when subjects were free to choose.

As previous public goods experiments predicted, the small amount of cooperation during the initial rounds in community N quickly eroded. As cooperation collapsed, players who initially chose community N in hopes of higher payoffs abandoned the community for the stable payoffs in community P. Indeed, the vast majority of players increase their contributions after making the jump from community N to community P.

The research of Gurek, Irlenbusch, and Rockenbach offers an important insight. When people are free to vote with their feet, we are much more likely to see the development of communities with strong cooperative norms, norms that persist even after people with previously weak cooperative norms join the community in an effort to benefit from its relatively high payoffs.