Difficulties in valuing street light improvement: trust, surprise and bound effects

Abstract

Previous empirical research has demonstrated an internal inconsistency that may occur in response strategies between the first and second valuations made to closed-ended contingent valuation questions. One possible explanation for this bound effect is the surprise of being asked the second valuation question, which may be enhanced where there is a lack of trust. This paper considers the use of closed-ended contingent valuation to estimate non-market benefits for an improved street lighting scheme where there is a lack of trust in the agency responsible for provision. The results provide confirmation that surprise is an important determinant of bound effects; however, efforts to reduce such bias using a prior statement of the bid range were found to be ineffective in increasing trust and reducing surprise. Given the importance of this area of research, directions for future research are considered.

Acknowledgements

This study was funded by the Department for Transport. Helpful comments were provided by Ian Holmes and David Reams at DfT. However, the views expressed in this article are those of the authors alone.

Notes

¹ The open-ended approach is more straightforward than the closed-ended approach as it merely asks the respondent for the maximum amount they would pay or minimum compensation they would accept in respect of the change in provision described. Although the open-ended alternative has advantages in terms of the ease of analysis, information provided and the absence of distributional assumptions, it has been much criticized, for example, in terms of incentives for strategic behaviour, sensitivity to the ‘fair-share heuristic’ and difficulty of the respondent task (Hoehn and Randall, 1987; Bohara et al., 1998).

² The open-ended approach can also be supplemented with a payment card, which may lead to more valid WTP responses (Donaldson et al., 1997).

³ The double bounded approach can also be extended to further bounds, however, Cooper and Hanemann (1995) and Scarpa and Bateman (2000) suggest further bounds provide little efficiency gain beyond that achieved by the first follow-up.

⁴ Calia and Strazzera (2000) considered the comparative efficiency of single and double bounded methods for different survey sizes and found the efficiency gains for small samples surveys to be particularly large.

⁵ For a summary see Bateman et al. (2001) and DeShazo (2002).

⁶ Based on the work of Kahneman and Tversky (1979), DeShazo (2002) uses prospect theory to explain bound effects through loss aversion and framing effects.

⁷ For example, if the bid range is £5 to £15, the middle value will be £10.

⁸ More recently, an approach has been developed to model zero values separately from those identified as protest responses (Strazzera et al., 2003).

⁹ Medians for the bootstrapped distributions were compared directly to see if they overlapped significantly. Difference of two means Z-test was performed to compare LBM estimates; however, the assumptions were in some cases violated and, where necessary, a comparison of distribution test was also performed for confirmation of the Z-test findings.