Choice Experiments: Application to Air Quality Policy Options and Investigation of Method's Incentive Compatibility

More than half of all Americans live in areas that violate at least one of the U.S. Environmental Protection Agency’s air quality standards for ground-level ozone and fine particulate matter. The majority of these areas have many options to come into compliance with the standards, such as requiring vehicle inspection/maintenance programs, changing gasoline blends, and requiring additional controls on power plants. Chapter 1 explores the use of conjoint-based choice experiments to assist local policymakers in determining the most beneficial policy strategy using the case study of Knoxville, Tennessee. Given that different policy actions pass costs onto households in different ways and that households may have varying preferences regarding how they pay those costs, I test whether willingness to pay for air quality improvements is sensitive to the method of payment. Potential heterogeneity of preferences is modeled through a mixed logit specification. Results indicate positive willingness to pay among Knoxville area residents for improvements in air quality, with vehicle inspection and an increase in the price of gasoline as the preferred payment vehicles over an increase in the electricity bill.

Chapter 2 investigates the underlying incentive compatibility of choice experiments with tests of the mechanism in an induced value laboratory setting. The theoretical properties of both dichotomous choice and trichotomous choice elicitation formats are explored under plurality voting rules and more generally under the assumption that the respondent perceives her decision to have some influence on the outcome. Results indicate that certain belief structures can lead to incentive compatible outcomes in a trichotomous choice format, depending on how a respondent believes the agency will incorporate respondent decisions into the provision of a public good. In addition, the trichotomous choice treatments had fewer deviations from theory and were less subject to status quo bias than treatments with dichotomous choice questions. Implications for the design of choice experiments and contingent valuation surveys are discussed.