Talbot Page

Voluntary Association in Public Goods Experiments: Reciprocity, Mimicry, and Efficiency

We find that a process of voluntary association where individuals express a preference about whom they want to be associated with can create strong incentives to increase efficiency and contributions in provision of a public good. This process of endogenous group formation perfectly sorted contributions by the order of group formation. Comparison of middle and last period behaviour suggests that a majority of the subject population are conditional cooperators, with a minority of monetary payoff maximisers. The experiment illustrates that under favourable conditions, where the opportunities of entry and exit are symmetrically balanced, a process of voluntary association can mitigate the free-rider problem.

Cooperation Under the Threat of Expulsion in a Public Goods Experiment

In a public goods experiment with the opportunity to vote to expel members of a group, we found that contributions rose to nearly 100% of endowments with significantly higher efficiency compared with a coexpulsion baseline. Expulsions were strictly of the lowest contributors, and there was an exceptionally strong fall-off in contributions in the last period, when the expulsion threat was unavailable. Our findings support the intuition that the threat of expulsion or ostracism is one device that helps groups to provide public goods.

Status Quo Bias in Bargaining: An Extension of the Myerson–Satterthwaite Theorem with an Application to the Coase Theorem☆

We use a generalized version of the Myerson-Satterthwaite theorem to study inefficiencies in bilateral bargaining over trade of an indivisible good, where there is two sided private information on the valuations. We show that when preferences are convex and quasi linear, and when the private information represents the magnitude of the utility gain or loss and follows a uniform distribution, that the most efficient mechanism always exhibits a bias towards the status quo. In the case that utility functions are quadratic in the amount traded, we prove that for any incentive compatible direct mechanism, there is an expected bias towards the disagreement point. In other words, for the class of preferences we study, there is a strategic advantage to property rights in the Coase bargaining setup in the presence of incomplete information.

An Experimental Study of the Effect of Private Information in the Coase Theorem

We investigate, in an experimental setting, the effect of private information on the Coase theorems predictions of efficiency and allocative neutrality. For a two-person bargaining game, we find significantly more inefficiency and allocative bias in the case of private information compared with the case of complete information. We also find substantial bargaining breakdown, which is not predicted by the Coase theorem. For the case of private information, we reject the Coase theorem in favor of the alternative of a generalized version of the Myerson–Satterthwaite theorem, which predicts inefficiency, allocative bias in the direction of the disagreement point, and some bargaining breakdown.

Taking the Coase Theorem Seriously

It is sometimes believed that technical apects of a theorem have little to do with the policy implications of the theorem. On the contrary, in this paper we argue that for the Coase Theorem, the technical details are very important in understanding the potential policy implications, since the two interact in a way that leads to a dilemma: a formally correct version of the theorem that yields the usual conclusions requires assumptions that are too restrictive to give the theorem much policy relevance. On the other hand, relaxing the assumptions of the theorem to be sufficiently plausible to be applicable in real world settings modifies the conclusions of the theorem.

Getting Punnishment Right: Do Costly Monitoring or Redustributive Punishment Help?

We introduce new treatments of a voluntary contribution mechanism with opportunities to punish, to see how contributions and punishments change when (a) each dollar lost in punishment must be awarded to another team member and/or when (b) obtaining information on individuals’ contributions is a costly choice. Conjectures that tying punishments to rewards might reduce punishment of high contributors (perverse punishment) or increase overall punishing are not completely born out, but innovation (a) nonetheless succeeds in making the net punishment of high contributors much less common because they receive enough rewards to offset punishment. A surprise finding is that innovation (b) also decreases the incidence of misdirected punishment, since high contributors do more monitoring than low ones while low contributors do most of the perverse punishing. Both innovations raise both contributions and earnings relative to the familiar VCM-with-punishment treatment.

Pivot mechanisms as a link between probability and preference revelation

This paper defines a pivot mechanism for revealing probability judgments and explores a parallel between pivot mechanisms in two Bayesian games: one for probability revelation and the other for preference revelation. The main result is that Bayesian equilibria must satisfy a simple fixed point property (which can be used constructively to find equilibria). The result provides simplification and extension of previous results for the Groves mechanism and kth price auctions. Further, pivot mechanisms generalize proper scoring rules, provide an analytic tool for optimizing research incentives, and under some conditions outperform proper scoring rules.

An Uninterpreted Spatial Version of the Trust Game: Evidence of Reciprocity Without Suggestive Words, Evidence of Iterated Dominance Self-Taught

In this working paper we report on two trust games: a BDM-like game which is interpreted through its use of the possibly suggestive words “show up fee,” “sends,” “tripled,” “send back”; and an uninterpreted spatial game that does not use these words suggestive or not. In the spatial game we found a considerable amount of reciprocity, which implies the words are not necessary for reciprocity. For further comparison we designed the two games to have a correspondence relation (the relation extends to the original BDM trust game). We focused on two “variables” – interpreted or uninterpreted and spatial or word-based. We also designed “constants” which were identical or near identical in the two games. We did this to reduce confounding in statistical comparisons. We found the frequency of reciprocity in the spatial game, without the suggestive words, was about the same as the frequency of reciprocity in the BDM-like game, with the suggestive words. We found iterated dominance in the spatial game was 5.5 times higher than in the BDM-like game. And we found sending the full endowment was significantly more frequent in the BDM-like game than in the spatial game.