Association of Environmental and Resource Economists

Shifting individual incentives is the core policy approach to environmental market imperfections. Yet its implementation involves some type of monitoring of either individual behaviors or their consequences. In some settings, that could be either physically impossible or excessively costly. For non-point-source pollution, such constraints on policy have led to a variety of ambient-based targeting mechanisms (Segerson 1988; Xepapadeas 1991). Here, we consider small-scale gold mining (SSGM ), which features such challenges for individual monitoring and enforcement but also constitutes an important economic option for many rural poor within the developing world. We examine one way states might effectively outsource some of the required monitoring, i.e., collective incentives. They reduce information burdens, yet they depend on group interactions. SSGM is said to involve ~15 million people globally. It is the world’s largest mercury polluter, putting 100 million people at risk of poisoning (Wade 2013). When transformed to organic form, mercury bioaccumulates in the food chain and the consumption of contaminated fish puts people at risk of both neurological damage and birth disabilities (National Research Council 2000).Yet mercury pollution is not SSGM's only environmental impact. Alluvial or hard-rock mining removes significant amounts of sand, gravel and rocks, contributing to both sedimentation and acidification of water sources given that SSGM utilizes large amounts of water and operations are usually close to water bodies so that emissions reach downstream populations (Shoko 2003). Mining also frequently is characterized by forest clearing and, at least for alluvial mining, the opening of pits that are filled with water and, thus, can become sources of diseases like malaria. Despite apparent inefficiency resulting from unregulated private choice within SSGM, the sector have not necessarily been considered sensibly in local legal frameworks and their applications. For instance, command-and-control measures that might be appropriate for state interactions with large companies have been suggested for regulating small miners − despite states’ obvious lack of ability to enforce such constraints upon scattered and small actors (Hinton 2005). This has yielded chaotic, open-access, unregulated and poorly organized mining (Shoko 2003) and thus it would seem that improved governance could reduce mining’s negative effects on others. Yet improved mining governance on the frontier is a challenge, given the costs of enforcement, and that much mining is of alluvial deposits adds the challenges of non-excludability in rivers. Few strategies exist for such settings. We explore an institutional innovation to regulate using a collective reward scheme conditioned on aggregate behavior easier to monitor than individuals. In our scheme, compliance is quasi-voluntary (Levi 1988) or induced by conditional cooperation. We designed an artefactual framed field experiment to test this targeting mechanism, drawing from literatures about moral hazard in groups (Holmstrom 1982), ambient-based mechanisms (Segerson 1988, Xepapadeas 1991) and threshold public goods (Cadsby and Mayness 1999). In our setup, the government sets the target for aggregate compliance, verifies the target and then provides a reward to the group conditional on collective compliance. The reward is split equally among miners. This limits the state's burden to verifying aggregates, leaving the community members to decide among themselves whether to reach the aggregate target and, if so, what each individual does. In principle, all pre-existing norms, social capital and prior interactions should affect the outcome. In addition, we can allow communication that may further trigger application of local institutions and norms. We recruited 600 participants in four mining communities within Colombia's Pacific region. We explore the effects upon compliance and the distribution of underlying individual behaviors of: policy severity (different aggregate targets); temporal ordering of severity (a sequence of aggregate targets is employed); and cheap talk (when individuals can talk before making decisions). We verify that such an instrument can induce groups to choose the socially chosen target, even in the case of higher aggregate targets when the collective reward barely outweighs the total costs. Nevertheless, even when aggregate behavior is very close to the socially set aggregate target, generally a fraction of individuals contributes less than the implied average, while others contribute more. This yields unequal payoffs. Even when, as expected, communication helps to solve coordination challenges, still some groups reach agreements involving unequal outcomes.