The regulation of nonpoint source water pollution from agriculture is a complex problem characterized by a multiplicity of polluters, informational asymmetries, complex fate and transport processes, and stochastic environmental factors. Taken together, these characteristics make regulatory policy based on individual firm emissions prohibitively costly. To circumvent this issue, economists, beginning with the seminal work of Segerson (1988), have devised economic incentive instruments that assign liabilities based on deviations between the observed ambient water quality level and a specified pollution threshold (Xepapadeas 1991; Horan, Shortle and Abler 1998, 2002; Hansen 1998, 2002). In the special case of a linear damage function, the regulator can optimally set the parameters of Segerson's (1988) incentive scheme solely with information on the damage function. When the damage function is nonlinear, a depiction that likely represents many watersheds, Segerson's incentive scheme is firm-specific, and the regulator must acquire costly firm-specific data on factors such as input use, land management practice, and soil type. Using a linear damage function setting, recent laboratory experimental economics efforts have investigated the ambient-based mechanisms proposed by Segerson, as well as some simple variants (Spraggon 2002, 2004; Poe et al. 2004; Vossler et al. 2005). A fundamental limitaion of this body of research, however, is that has utilized an "optimal design" in which the threshold pollution level for triggering the abient-based policy is set equal to the social optimum. It is therefore unclear whether subjects are optimally responding to the tax and threshold combination, or simply trying to reacting to the focal point created by the threshold. A second limitation of past experimetnal economics research is that, following Segerson, these investigations have utilized the limited case of a linear tax function. While a tax policy is relatively straightforward to apply when damages are linear, the application to real world situations may be limited. A more believable circumstance is that economic damages increase at an increasing rate as ambient pollution levels rise. This paper advances the experimental literature on ambient based pollution mechanisms in two important ways. First, by employing a range of marginal tax rates and threshold levels, we show that subjects do in fact respond optimally to the tax and cutoff combination. Second, by using the damage based tax proposed Hansen (1998) and Horan et al. (1998), we show that aggregate results when the economic damages from ambient water pollution are nonlinear are not significantly different from corresponding results under the linear tax.
