Public goods experiments and a theoretical treatment of efficient bilateral network

Abstract

This dissertation contributes to this theoretical and experimental literature by identifying three settings where efficiency can be improved. By conducting a local and global public goods experiment, Chapter 2 studies the effects of endowment heterogeneity by comparing a setting where two localities have the same income (homogeneous treatment) with a setting where the localities differ in income (heterogeneous treatment). We find that: 1) social efficiency is lower in the heterogeneous treatment; 2) the efficiency difference comes from two aspects: a shift from global to local public good contribution and a lower total public goods provision; and 3) inequality aversion and reciprocity both play a role in contribution behavior; however the former does not fully counter efficiency reduction, and the latter manifests an undesirable side effect: people reduce global public goods contributions to reciprocate on local public good contributions. Chapter 3 explores the causal mechanisms through which real-time information affects energy consumption by conducting a randomized-control trial with residential households. The experiment disentangles two competing mechanisms: (i) learning about energy use via real-time feedback is sufficient to sustain energy conservation, or (ii) having a constant reminder of energy use is necessary for energy conservation. We have two main results. First, we find a statistically significant treatment effect from receiving an in-home display. Second, we find that learning plays a more prominent role than saliency in driving energy conservation. This finding supports the use of energy conservation programs that enhance consumers' learning about energy use. Chapter 4 studies mechanisms that incentivize agents to contribute their time to the level that generates the maximal profit for the society at the Nash equilibrium (we call this efficiency). We focus on the case of bilateral projects, that is when every project requires time allocation only from two agents. Our main result is the characterization of all the mechanisms that satisfy efficiency for any set of production functions. Furthermore, we characterize efficient mechanisms that satisfy other desirable properties such as monotonicity in the payoffs, strong Nash equilibrium of the efficient outcome, or consistency in the number of participating agents.