Despite calls for more integrated residual management, research on the performance of methods to regulate pollution has paid little attention to cases of multiple pollutant control. This study helps rectify this omission in multiple ways. In Chapter 1 I describe the many issues that arise in the control of multiple pollutants and identify where economics and policy have addressed them. In Chapter 2 I extend a well-known proof demonstrating that emissions taxes or cap-and-trade instruments may yield a Pareto optimal outcome in a general equilibrium setting to the case where there are multiple pollutants in the economy. Chapter 2 also includes an exploration of how changing the joint abatement relationship in a deterministic firm-level model affects emissions and allowance prices when taxes and/or cap-and-trade programs are used.

Chapter 3 extends a model commonly used in the instrument choice under uncertainty literature to the case of jointly abated pollutants. In the single pollutant case with uncertain abatement costs, Weitzman (1974) and others have demonstrated that the expected welfare from an emissions tax is likely not the same as for a tradable emissions cap and derive conditions under which each instrument is preferred to the other. I find that the criteria identifying the welfare-maximizing instrument in the single pollutant framework may be misleading in identifying the optimal set of instruments in a multiple pollutant framework. I also show that the optimal instrument for any one pollutant may depend on how the other pollutants are controlled.

A case study is then explored in Chapter 4. A market simulation model of the national electricity sector is employed to determine the optimal instrument combination to control sulfur dioxide and mercury emissions from coal-fired power plants. There are important, but uncertain, relationships in the abatement of these pollutants. The analysis shows that the optimal instrument combination consists of controlling mercury by a tax and sulfur dioxide by a tax. The cost of selecting a suboptimal instrument mix is between $90 and $190 million (2004 $) in 2020 depending on which suboptimal mix is adopted. Chapter 5 summarizes the findings and suggests issues for future research.