Essays on environmental policy design for agriculture under multifunctionality, spatial variability, and risk

Environmental policies in agriculture are limited by five problems: their relationship to trade, unobservable effluents, multifunctionality, spatial variability, and risk. Policy design is complicated further because several of these problems occur jointly. While the existing literature approaches the problems individually, the essays in this dissertation develop policies under multiple complications, which incorporate potential feedback effects.;The first essay designs policies at the national level, accounting for unobservability, multi functionality, and trade. Amenity benefits from farmland are protected through an acreage subsidy, while agricultural pollution is controlled indirectly through a tax on polluting inputs. Optimal joint policies are derived in a general equilibrium setting for closed and open economies, where large countries exploit terms-of-trade effects. Based on a simulation of U.S. agriculture, the relationship between policies is empirically important. The farmland subsidy exceeds its amenity value by about 50%, and the joint policies distort world prices by an estimated 9%.;In the second essay, the joint effects of unobservability, multifunctionality, and systematic spatial differences are identified in the context of suburban sprawl. Farmland generates amenity benefits as well as pollution, but the value of these externalities depends on their location relative to suburban households. Optimal taxes on development and polluting inputs are derived in a dynamic model of land conversion, and are applied empirically to a typical metropolitan area in the U.S. The taxes influence the timing of development in opposite directions; the optimal joint policies lead to substantially less pollution and a slightly faster rate of conversion.;The third essay designs policies that allow for unobservability, risk, and unpredictable spatial variation. To overcome the government's lack of information on each farmer's technology and risk response, the policy allows farmers to choose from a set of regulations. The central question is whether a payment can be attached to each regulation so that every risk-averse farmer selects one appropriate for his farm. Based on second-degree stochastic dominance, a set of empirically testable conditions is derived to identify the situations where self-selection is possible. These conditions, and an associated empirical procedure for finding payments, are applied to the nitrate contamination problem in New York.