Groundwater games: Users' behavior in common-pool resource economic laboratory and field experiments.

다국어 초록 (Multilingual Abstract)

Groundwater currently provides 30 percent of freshwater in the world, and it is estimated that it potentially constitutes approximately 89 percent of the world's fresh water. Nevertheless, extensive extraction of groundwater in some areas is leading to aquifer depletion.
Groundwater is considered, in some situations, a common-pool resource (CPR) with extremely high use value. Economists and political scientists have devoted a great amount of effort to understand how common pool resources are managed, as well as the characteristics of institutions that emerge in order to deal with the use and distribution of CPR. The characteristics of a CPR and how agents interact in their use are important for the success and sustainability of CPR-dependent communities.
This dissertation will analyze the behavior and decision rules of CPR users in an experimental context. We discuss the results from laboratory and field experiments framed as a dynamic groundwater game in which users pump water from a shared resource, and the actions of users in one period affect resource availability and extraction costs of everyone in the next period. This is an innovative design, since most of previous studies do not consider the groundwater problem in experimental settings, and do not conduct a groundwater experiment with actual groundwater users.
The objectives of this research are twofold. In the first part of the dissertation, I assess the presence of strategic behavior when participants have access to an efficient technology that yields both private and public benefits to CPR users. This situation might discourage users from adopting the new technology, which results in free-riding. In this part of the dissertation, I am also concerned about the effectiveness of group arrangements to improve water usage and to guarantee appropriation levels that yield a socially-desirable economic outcome. Given the experimental design, I can partially address both learning effects during the game and unobserved heterogeneity of participants, in order to obtain a precise estimation of the impact of the treatments. Finally, I also analyze the factors that affect technology adoption and deviation from agreed water usage.
In the second part of the dissertation, I address the degree of heterogeneity in actions of participants, and aim to identify different types of behavior among participants in the use of the shared resource . Using a methodology proposed by Geweke and Keane (2000) and Houser et al. (2004), I can relax the rational expectations assumption in the dynamic choice problem, and identify different behaviors in the population. Based on the decisions that participants make during the experiment, I can identify and cluster parameters that define the future function of the dynamic choice problem in different "types" or groups of participants, with these groups endogenously created. I propose a flexible mixed multinomial discrete choice model that allows parameters to be drawn from a discrete mixture of distributions. In order to estimate the parameters and group membership probabilities, Bayesian Markov Chain Monte Carlo techniques are used. These methods allow the estimation of highly dimensional discrete choice problems and adds flexibility to estimation and the clustering process.