Specific Information in the Contingent Valuation of Environmental Risks: The Case of Nitrates in Groundwater

Information, or lack thereof, is an important input in value formation and the distribution of contingent values. Although most conceptualizations of the contingent valuation process stress that information provision should be 'adequate,' very little empirical research has been devoted to assessing the effects of different information flows on contingent values and to establishing a standard of information adequacy for contingent valuation studies. The need for such research is particularly cogent for valuing the benefits of reducing environmental risks, a non-market good which is increasingly being valued with the contingent valuation method. Using nitrates in groundwater as a case study, this paper evaluates and compares health risk perceptions and the distribution of contingent values for groundwater protection associated with three different levels of information provision. In the first level, no information about nitrates or personal exposure was provided to the participants, an approach which reflects the philosophy that values should be based on prior information and preferences. The second information level provided participants with general information about the health effects of nitrates, sources of nitrate contamination, government standards for nitrates, indicators of the distribution of nitrate levels in local wells, and opportunities for averting behavior. The third level of information flow provided the general information packet along with specific information about actual nitrate levels found in participants' wells. The primary results of this research are that individuals need afull-information set that includes both general and specific information to identify their own best interests with respect to groundwater protection programs, and that the provision of general information alone appears to lead to biased estimates of willingness to pay for groundwater protection. These results establish a full-information standard for future contingent valuation research of groundwater protection. • The authors are, respectively, Assistant Professor, Department of Agricultural Economics, Cornell University, and Professor, Department of Agricullural Economics, University of Wisconsin-Madison. This research and paper were funded, in part, by the Center for Integrated Agricultural Systems, University of Wisconsin-Madison, and the College of Agriculture and Life Sciences, Cornell University. The authors are indebted to Jean-Paul Chavas and Bill Provencher for comments on the research, and to Patricia Champ and Dan Mullarkey for their help with the data collection. ­ This paper was presented at the Association of Environmental and Resource Economists sponsored .. session of the Annual Meetings of the American Agricultural Economics Association, Orlando, August 1993. Prior Infonnation, Genend Infonnation, and Specific Infonnation in the Contingent Valuation of Environmental Risks: The Case of Nitnltes In Groundwater Information is an important input in value formation and the distribution of contingent values [Hoehn and Randall; Bergstrom and Stoll; Samples, Dixon and Gowen; Boyle; Bergstrom, Stoll and Randall]. Although critical assessments of the contingent valuation process stress that information provision should be 'adequate' in order to obtain satisfactory transactions and reliable values [Fischhoff and Furby; NOAA], very little empirical research has been devoted to establishing a minimum standard of information adequacy for contingent valuation studies. The need for such research is particularly cogent for valuing the benefits of reducing environmental risks such as groundwater contamination, as this is an unfamiliar commodity for most households and previous research indicates that risk perceptions are affected by new information [Viscusi and O'Connor; Smith et al.; Smith and Johnson]. Using nitrates in groundwater as a case study, this paper evaluates health risk perceptions and the distribution of contingent values for groundwater protection associated with different levels of information provision. The evidence presented in this analysis suggests that general information about nitrates, specific information about exposure levels, and prior information affect contingent values, and that individuals update their perceptions of groundwater safety with new information. Evaluations of individuals' abilities to assess their reference and target risks associated with a nitrate protection program suggest a fullinformation standard that includes both general and specific information for future contingent • valuation research of groundwater protection. 2 Conceptual Framework The conceptual framework underlying this analysis is based on the ex ante statistical life model in which exogenous risks are modified by self-protection activities [see Berger et al., Shogren and Crocker]. In this framework it is hypothesized that individuals optimally select exposure and averting activities in order to maximize state dependent expected utility. The extension of this framework to willingness to pay for groundwater protection from nitrates requires that the consumption and averting activities be specified. Here we assume that households consume drinking and cooking water (X) from three sources (i), each with an associated price (p) and nitrate level (N). These include: water consumed from the household well (X1(N1»; water consumed from sources outside the home (XiN2» such as at school, restaurants, neighbors' homes etc.; and sources of water consumption intended to mitigate exposure (X3(N3», including the installation of purification systems, importation of water from 'pure' wells, and the purchase of bottled water. Let Xdenote the vector of water consumption, N denote the associated nitrate levels, p denote the vector of prices, and define Px = ptXI(N) + P2X2(N) + P3X3(N). For simplicity assume that N3=O < Nt,N2. Uncertainty is introduced into the model at two levels. Because of the stochastic nature of biological and physical transport, it is reasonable to define Nt and N2 as random variables with a joint probability of 11 =11(Nt ,N2). Second, under the state dependency framework of the statistical life theory, health outcomes play an important role. Yet, given exposure to a hazard, future health outcomes (h) remain a random variable. Let this uncertainty be characterized by the conditional probability density function f(h;x'11) and let F(h;x'11) represent the associated cumulative distribution function defined over the set 3 of possible health states A. The distribution of anticipated exposure levels and health risks are subjective and infonnation dependent, implying that infonnation levels need to be explicitly identified in the model. In specifying infonnation levels, our analysis distinguishes between infonnation about nitrates that is general in nature and infonnation that is specific to a household's exposure level from its own water source. General information (GI) about nitrates would thus include possible health effects and sources of nitrates, government standards, and opportunities for mitigation. With this bundle of infonnation, the decision maker could conceivably define health effects and optimal averting and consumption strategies for each hypothetical level of exposure. Specific information (SI) about nitrate levels found in an individual's well would affect the subjective distribution of nitrate exposure levels. On this basis, general and specific infonnation can be incorporated into the subjective distributions as follows. Joint distributions of nitrate exposure levels are treated as a function of both general and specific infonnation