Consider the Hanemann and Cameron specifications shown in examples 19.3 and 19.4.

(a) For the full dataset, summarize the proportion of people who vote “yes” in the referendum for each of the four bid levels. Does the proportion of yes votes respond to bid levels in the way you would expect?

(b) Replicate the example 19.3 and 19.4 estimates by running the specifications shown in the chapter, using only the individuals who responded to version I. Note that the predictions for willingness to pay are random variables since they are functions of estimated parameters. Predict the standard errors and confidence intervals for the welfare measures, using a bootstrap or alternative strategy.

(c) Run the same models using only the individuals who responded to version III.
Comment on any differences in the estimates you recover for the value of the program.

(d) Now run models that pool observations from version I and version III. Modify your specifications so that you are able to distinguish differences in the willingness to pay for programs I and III. Do you find a statistically significant difference? What do your findings suggest about sensitivity to scope?

(e) Survey respondents answered a follow-on question in which they reported the level of confidence they had in their answers to the referendum questions. Based on this, individuals are classified as very certain, certain, or uncertain. Repeat part (c), but this time drop respondents who report they are uncertain. Does this change your findings on sensitivity to scope? Rather than dropping uncertain respondents, change their votes to “no” (if they voted yes). Explain how this changes your value estimates and findings on scope.

Example 19.3

Example 19.4

Transcribed Image Text:

In the Phaneuf et al. water quality study, 302 people from the southeastem US answered the version of the survey that included the question shown in example 19.1 ; 55 percent answered "yes" to the bid amount they were presented with. Among the respondents there were 81 people who were presented with B=$24, of whom 73 percent answered yes. This falls to 51 percent, 50 percent, and 44 percent