The effect of signalling and beliefs on the voluntary provision of public goods

We report and analyze the results of a series of classroom experiments on the voluntary provision of public goods. Using fixed effect panel regression models we find that cooperation significantly increases when participants are forced to guess the degree of overall cooperation. We also find that the possibility to make announcements during the experiment enhances cooperation significantly although these commitments are not binding and heavily used for cheating purposes. We suggest that this effect is due to attempts at restoring the announcements' reputation in the aftermath of heavy cheating.