A New Approach to Handling Uncertainty in the Optimal Design of Environmental Policy. Nii Adote Abrahams, Department of Agricultural Economics, Pennsylvania State University, University Park, PA 16802
Even though it is clear from the work of Weitzman and others (e.g Stavins) that uncertainty is a critical factor in policy choices, there is little empirical analysis of the implications of public uncertainty about benefits and costs for the choice of policy instruments. Using a simulation model of aggregate U.S. corn production, the performance of alternative policies under public uncertainty about the costs and benefits of nitrate pollution control in the production of corn are compared. Policies are chosen to maximize expected net benefits. Public uncertainty is modeled as uncertainty by regulators about such economic parameters as demand and input substitution elasticities, and the environmental costs of nitrate pollution, and also uncertainty about environmental parameters, such as nitrate losses and delivery. Uncertain parameters are treated as random variables with known distributions. Because net benefits cannot be obtained as a closed form expression of exogenous variables, a numerical procedure is used to solve for the endogenous variables for any given set of policy and exogenous parameter values. Parameter distributions are approximated with discrete points and associated probability values - that is, a discrete distribution - using the Guassian Quadrature approach to numerical integration. A search procedure is then used to determine the optimal policy. All of these issues are considered with and without traditional government commodity programs that distort input and output markets. Results indicate that economically efficient nitrate policy control in the production of corn are compared. Policies are chosen to maximize expected net benefits. Public uncertainty is modeled as uncertainty by regulators about such economic parameters as demand and input . . . . . . [RiskWorld Note: Submitted abstract incomplete]