Essays on the measurement of environmental externalities

This thesis is a methodological, empirical and theoretical investigation of issues that arise in the measurement of environmental externalities, using stated preference approaches. Chapter 1 introduces the methodologies to be used: the contingent valuation and the contingent ranking methods. Chapter 2 shows how contingent valuation surveys may have to be modified for implementation in developing countries. A contingent valuation survey is used to elicit the compensation required by Peruvian Amazon farmers to supply carbon sequestration services, by switching from slash-and-bum agriculture to forest preservation and agroforestry. The results show that participatory market-based approaches could enhance the effectiveness of traditional forest conservation efforts, while benefiting resource-poor farmers. Chapter 3 applies a double-hurdle approach to model willingness to pay for cultural heritage protection, in the presence of a large proportion of people stating a zero willingness to pay, using data from a contingent valuation study of Bulgarian monasteries. In addition, the chapter explores a number of financing mechanisms that can be used to turn willingness to pay into actual cash flows in the presence of a heterogeneous value distribution. Chapter 4 considers the suitability of a family of models for modelling the non-use value produced by programmes for preserving endangered species. The models are tested by considering how well they perform in the context of a valuation exercise where the value estimate for one preservation programme must be transferred to another similar programme. The results of a contingent valuation experiment indicate that the standard model for valuing such programmes does not perform well. Models involving some but not all types of option value are found to perform adequately. Models involving non-standard discounting are also considered. Existing best practice guidelines for conducting contingent valuation have evolved from the experience of applications in developed market economies. Transition economies pose specific kinds of problems such as low incomes, political instability, and lack of experience with market mechanisms. There is an issue to whether valuation methods based on hypothetical markets can successfully be used for informing policy decisions in this distinct framework. Chapter 5 reports the results of the first large-scale contingent valuation study conducted in Hungary to evaluate the benefits of water quality improvements at Lake Balaton. Chapter 6 uses the contingent ranking method to estimate the value of the human health and bio-diversity impacts associated with pesticide applications, using a 'green' consumer product as a payment vehicle. The resulting estimates, which perform well in terms of standard validity tests, are subsequently used to shed light on the design of an 'environmentally friendly' bread product and on the structure of a possible pesticide tax. Chapter 7 develops a set of simple tests for the logical consistency of responses to contingent ranking surveys based on the concepts of dominance, rank-consistency and transitivity of rank order. The findings are that, even in a relatively simple experimental design, nearly half of respondents fail to provide fully coherent responses. The presence of inconsistent rankings is shown to have a statistically significant effect on the coefficient estimates obtained from standard econometric analysis of the data. However, the problem becomes less severe when welfare measures are calculated from the combined data set of consistent and inconsistent responses. Finally, Chapter 8 undertakes a systematic comparison of the contingent valuation and the contingent ranking methods, and investigates the extent to which either of them is affected by part-whole bias. The results indicate that contingent ranking values exhibit much greater sensitivity to scope than contingent valuation values. Contingent ranking gives significantly larger results than contingent valuation for the more inclusive public good and significantly smaller results for the less inclusive public good. In general, these results are highly robust to the statistical specification chosen to model either the valuation or the ranking data. (Abstract shortened by UMI.