Part-whole bias in contingent valuation: will scope effects be detected with inexpensive survey methods?

• Author: Whitehead, John C.

1. Introduction

   There is currently much interest in measuring the benefits of environmental quality improvements for policy purposes. An ongoing controversy involves the use of the contingent valuation (CV) method to estimate the total economic value, including non-use value, for environmental resources damaged by commercial activities (Hausman 1993; Carson et al. 1994; Portney 1994; Kopp and Pease 1997). The CV method is a survey approach to the valuation of resource allocation changes. Survey respondents are directly asked about their willingness to pay for increments or to avoid decrements in the environmental resource. Nonuse value is the lost consumer surplus from economic activities (e.g., an oil spill) suffered by a household that has never consumed the resource on-site (e.g., recreation). The measurement of total economic value is problematic since a positive nonuse value does not require any associated behavior (Smith 1987). Economic values for clean environments are typically measured through observation of recreation, averting, and migration behavior (Freeman 1993). People who do not participate in recreation or other activities while enjoying the knowledge of the existence of a natural resource do not provide the behavior necessary to measure these values with conventional methods. The CV method does allow measurement of the nonuse values.

   A major concern with using CV to measure total economic values that include nonuse values is whether these measures are consistent with economic theory (Diamond and Hausman 1994; Hanemann 1994). One issue that has received considerable attention is whether CV results pass a "scope test" (Arrow et al. 1993). Mitchell and Carson (1989, pp. 250-1) define part-whole bias as one type of respondent insensitivity to scope:

   Part-whole biases are major amenity misspecifications, and are also a result of the tendency of respondents to respond to public goods as global symbols without paying sufficient attention to the specific description offered in a CV survey. . . . The dimensions of a good that are particularly prone to this misperception are its geographic distribution, its benefit composition, and the package of policies of which it is a part.

   Through the implementation of carefully designed surveys, researchers should manage to elicit total economic value, or willingness to pay (WTP), which does not decrease with increases in the quantity or quality of the affected environmental resource. If CV-based total economic values are determined to be valid and reliable measures of economic welfare, the conduct of environmental policy analysis might undergo significant change (Portney 1994).

   Some recent empirical evidence on the ability of CV to detect scope effects is negative.(1) Kahneman and Knetsch (1992) and Fischhoff et al. (1993) do not find scope effects when using open-ended WTP questions in telephone surveys. Boyle et al. (1994), using a mall-intercept survey, and McFadden (1994), using a telephone survey, do not find scope effects using open- and closed-ended WTP questions.(2) On the other hand, several studies have found that WTP is sensitive to the scope of the resource change while employing in-person interviews and mail surveys (Loomis, Lockwood, and DeLacy 1993; Carson and Mitchell 1995; Carson et al. 1996).(3) This comparison of studies suggests that it may be difficult to find significant scope results without survey methods that allow the use of visual aids.(4)

   The research that has found respondent insensitivity to scope has been criticized for using relatively inexpensive survey methods, not using visual aids, and eliciting WTP with open-ended questions (Smith 1992; Carson and Mitchell 1993, 1995; Loomis, Lockwood, and DeLacy 1993). However, Schuman (1996) discusses the benefits of mail and telephone surveys, including cost, and asserts that they are useful for CV research, especially experimental treatments.(5) The results of Carson and Mitchell (1993) and Berrens, Ganderton, and Silva (1996) suggest that insensitivity to scope may not be a result of the use of inexpensive survey methods but rather induced directly by survey design or the type of environmental goods considered.

   In this paper, we conduct scope tests with a split-sample survey that elicits both use and nonuse values. We employ standard telephone survey methods with no photographs or other visual aids. The WTP is elicited with closed-ended questions. Respondents are presented with short descriptions of the environmental resources, policy descriptions, payment rule, and payment vehicle. We first sketch the theory and describe the sample for our application: valuing water quality improvements in the Albemarle and Pamlico Sounds in North Carolina. We then describe the data and present the results of the scope test. Our conclusions follow.

2. Economic Theory

   Suppose that survey respondents possess a utility function, u([x.sub.1], [x.sub.2], [q.sub.1], [q.sub.2], z), where [x.sub.i] is the number of recreation trips to site i (i = 1, 2), [q.sub.i] is quality at recreation site i, and z is a composite commodity of all other goods. Recreation site 1 is the Albemarle Sound, and site 2 is the Pamlico Sound. The expenditure function, e([p.sub.1], [p.sub.2], [q.sub.1], [q.sub.2], u), is obtained by minimizing expenditures,

   m = [summation of] [p.sub.i][x.sub.i] where i = 1 to 2 + z,

   subject to the utility constraint, where m is income, [p.sub.i] is the price to site i, and both are normalized by the price of z. Willingness to pay for an improvement in Pamlico Sound quality is

   [Mathematical Expression Omitted],

   where [Mathematical Expression Omitted]. Substitution of the indirect utility function evaluated at improved quality, [Mathematical Expression Omitted], into the expenditure functions yields the variation function, [s.sub.i]([center dot]):

   [Mathematical Expression Omitted],

   which is increasing in income, decreasing in own-price, either increasing or decreasing in cross-price, and increasing in own-quality (Whitehead 1995).

   ...