|
Abstract:
|
Policies designed to reduce pollution-related health burden by limiting harmful emissions from US power plants are highly contentious and in need of supporting evidence. Evaluating these interventions is met with two key challenges. First, interventions are implemented at power plants, but key questions for regulatory policy pertain to how emissions reductions unfold to affect pollution and health outcomes across the country. Thus, the units at which the interventions are defined and implemented (power plants) differ from the units at which outcomes defined and measured (residential locations or individuals). Second, pollution exposure and health outcomes at a given location are dependent upon interventions applied at many power plants, which is known in the causal inference literature as interference. Collectively, we term the setting of causal inference with interference among two levels of observational unit bipartite causal inference with interference. We outline the development of new statistical methods for this setting in the context of evaluating the effectiveness of policies impacting power plants across the US.
|
