Decisiveness, attitude expression and symbolic responses in contingent valuation surveys¹

Abstract

Despite suggestions that respondents to contingent valuation studies often follow alternate models of individual decision making, few theoretical models have been developed. The paper presents a theoretical model of symbolic and attitude-expressive contingent valuation responses, drawing on relevant contributions in psychology and political science. It is argued that the highly symbolic and emotional nature of many environmental issues often activates the need for individuals to express their attitudes and values, which when coupled with a perceived non-decisiveness of individual questionnaire responses, can result in value-expressive considerations dominating some CVM responses at the expense of the desired outcome-appraisal economic tradeoffs. The implications for some topical issues in CVM are explored.

Introduction

Economists are divided on the usefulness of Contingent Valuation, CVM, as a source of information on the value of environmental public goods for input to cost benefit analysis, CBA. While some see the method as inherently flawed, many would argue that it is already producing useful results and can be further improved by research directed at improved survey design etc. The exchange between Kahneman and Knetsch, 1992a, Kahneman and Knetsch, 1992b, Smith, 1992and Harrison (1992)provides a vivid illustration of the controversial nature of the technique.

Since many of the major criticisms of CVM relate in one way or another to how individuals respond to CVM questionnaires, one might expect that considerable work might be directed toward better understanding the motivations that lie behind CVM responses. Although some authors such as Edwards (1986)and Stevens et al. (1991)have called for more research into such motivations, there is to date a dearth of evidence on this topic. There is also rather little theoretical work on respondent motivations. This paper addresses these areas of inquiry. A model of CVM respondent behaviour is presented which involves the expression of symbolic attitudes in an institutional context where it may be rational to discount instrumental considerations. It is argued that such a model can better explain some observed features of CVM responses than some currently favoured explanations.

The notion of symbolic CVM responses dates at least to Kahneman (1986), who found that the proportion of respondents willing to pay various amounts of money to clean up lakes was `strikingly similar' for the Haliburton region of Ontario, the Muskoka region of Ontario, and all of Ontario. Kahneman attributed this `embedding' type effect to ideological contamination that results in symbolic demand. He noted that: `people seemed to be ready with an answer before the relevant numbers are specified' (p.190), and stated:

• The dollar number merely expresses the strength of the feeling that is aroused by these questions. Because the questions all elicit symbolic expressions of the same attitude, there is not much difference between the numbers that are attached to a single region and to all of Ontario...I call this `symbolic demand' because it is true of symbols that quantity is sometimes irrelevant: a small lake can be as good a symbol as a large one.

The model presented in this paper extends the work of Kahneman (1986)by linking relevant contributions in psychology and political science, and applying them in the context of current state-of-the-art CVM practice. The paper is organised as follows. Some background in terms of the literature, mainly in social psychology, on symbolic attitudes is presented in Section 2. In Section 3, CVM respondent behaviour is considered within a decision theoretic framework which allows for utility derived from the expression of symbolic attitudes, and which allows for perceptions of survey impact on outcomes. A simulation model is used to explore the empirical implications of the model. Section 4is concerned with broadening the model presented in Section 3to include biases associated with implementation of payment vehicle, and Section 5considers the phenomenon of `sending a message'. The implications of the model for some current issues in CVM research are explored in Section 6, and some conclusions are presented in Section 7.

A dichotomous-choice referendum format is assumed in this paper, as advocated by Mitchell and Carson (1989)and Hoehn and Randall (1987). Following this format, respondents are asked if they would vote at a referendum in favour of a specified environmental improvement, q¹–q⁰, if there was a cost $c to each individual associated with this outcome. By varying the $c amount across different subsamples, estimates of mean and/or median individual willingness to pay (WTP) can be obtained, and aggregated over the population of interest, to obtain an estimate of total economic value for subsequent inclusion in a cost benefit analysis. Focussing on this particular format facilitates the discussion, and is perhaps the easiest with which to draw comparisons with electoral choice contexts, which are the origin of the model developed in Section 3. It can also be noted that this format is increasingly seen as the most suitable for the majority of CVM applications for reasons of face validity, incentive compatibility etc. [Mitchell and Carson, 1989, Hoehn and Randall, 1987].