Introduction to Attribute - Based Stated Choice Methods

Several major environmental statutes enacted in the 1970s designate resource management agencies as trustees of the natural resources on behalf of the public, and enable the trustees to recover damages for injuries to public resources from releases of hazardous substances and discharges of oil. The standard measure of damages in the various statutes is the cost of restoring the resources to baseline conditions (" primary restoration ") plus compensation for the interim loss of resources from the time of the incident until full recovery from the injuries. 1 Though the measure of damages for interim losses was originally characterized in terms of monetary compensation, 2 trustees are only allowed to spend their damage recoveries on enhancing or creating (" restoring, rehabilitating, replacing or acquiring the equivalent of ") natural resources. The statutory restriction on the use of the recoveries has motivated the development of an alternative resource compensation measure of damages for interim losses-the cost of " compensatory restoration " actions that provide sufficient resources and services to compensate for the losses. The National Oceanic and Atmospheric Administration (NOAA) incorporated the resource compensation measure in its 1996 regulations implementing the natural resource liability provisions of OPA. 3 Under the new rule, the focus of the analysis becomes identifying and evaluating the attributes of environmental resources (and the services they provide) at both the injury and replacement resource sites. Current methods of environmental valuation can be categorized as Revealed Preference methods (RP) and Stated Preference methods (SP). RP methods (such as travel cost or hedonic price models) can be used to construct resource compensation profiles to a certain extent. For example, random utility models based on RP information can be used to determine how attributes from some alternatives in the choice set can be used to compensate for damages to attributes in another alternative. However, RP data can be found wanting for a number of different reasons subsequently reviewed, not least of which is that the behavior of interest may not be observable. Furthermore, RP data may suffer from a variety of problems 1 Throughout this discussion we assume that compensation is determined by evaluating the appropriate individual measures of welfare (compensating or equivalent variation). Clearly, this is a specific normative approach that is not without controversy. In practice this approach almost always ignores interactions between agents or interdependent preferences which may be very important in the case of public goods …