Fooks, Jacob R.Messer, Kent D.Suter, Jordan F.2018-01-302018-01-302018-01http://udspace.udel.edu/handle/19716/22638This research uses experiments involving student and farmer participants to explore the informational structure of nonpoint source pollution policies in a spatial, physically realistic stream setting with high frequency end of stream sensing. The data allows for an approximate solution to the “attribution” problem using an artificial neural network based recursive filter. This provides estimates of individual parcel contributions to the ambient pollution level. These estimates are noisy, and prove to have an interesting endogenous error structure. This is incorporated into a policy which imposes differentiated taxes on polluters. Laboratory experiments compare this estimated pollution source policy, ambient exogenous targeted tax policy, and a perfect information policy interacted with corresponding information sets. After taking part in all treatments, participants’ values for these different policies were measured using a voting mechanism. Additional information on the part of the regulator leads to increased production over the efficient level, with additional participant information exacerbating the problem. Even though the ambient policies performed better in aggregate, the voting data indicated positive willingness-to-pay for the estimated policy by some students who were located to strategically benefit from this information, while farmer subjects had a large positive willingness-to-pay for exact information, irrespective of parcel location.en-USNonpoint source polutionExperimental economicsNeural networksWorking Paper