The good of the few: Reciprocal acts and the provision of a public bad

• Favor-trading in public bad provision may occur in cases of logrolling or corruption.
• Subjects in a lab experiment trade favors when they have the information to do so.
• Favor-trading could increase public bad provision, but does not in our results.
• A ready-made coalition enables more reciprocation, including a pro-social channel.
• We show that reciprocity and information can have anti-social effects.

People trade favors when doing so increases efficiency. Will they when it reduces efficiency, such as in political logrolling? We introduce the “Stakeholder Public Bad” game, in which common fund contributions increase one person's earnings (the “Stakeholder”) while reducing others’ earnings and overall efficiency. The Stakeholder position rotates through all group members or just alternates among two people (making it easier to form a coalition). High Stakeholder rewards provide a lever for reciprocity: if someone contributes when another is Stakeholder, he may be rewarded with a gift when he becomes Stakeholder. Reciprocity is only possible when agents know others’ roles and actions, so information provision may be pro- or anti-social. In a laboratory experiment, we combine a rotating or alternating asymmetry in payoffs with varying levels of information provision to examine pro- and anti-social reciprocity. We find that subjects in the Stakeholder role willingly sacrifice social welfare. We also see both anti-social and pro-social favor trading, particularly when coalition-forming is easier. Favor trading does not change the mean level of public bad provision. People who trade favors tend to be less risk averse.