RANDOM UTILITY MODELS:

BACK TO OUTLINE

Random Utility Models

Kaoru, Yoshiaki; Smith, V. Kerry; Liu, Jin Long, "Using Random Utility Models to Estimate the Recreational Value of Estuarine Resources," American Journal of Agricultural Economics, 77(1), February 1995, 141-51.

1. This is a two-equation model. What are the two equations? Why is the second one necessary?
2. RUM models rely on pre-defined choice sets. Is this a problem in environmental valuation for recreational resources?
3. What forms of site aggregation have been used in the past?
4. What is the main limitation of the Parsons and Kealy strategy of using randomly selected alternative sites?
5. How doe the three specifications in this paper differ? What are their iimplications for the coefficient on travel costs? For the resulting benefits measures?
6. Briefly evaluate the quality of the typical RUM choice data used. Summarize the physical environmental quality data.
7. How is the total catch model specified? Why?
8. How were explanatory variables calculated for sets of aggregated sites?
9. Does this paper support Parsons and Needelman's findings that benefit estimates are sensitive to aggregation level?
10. What is the problem with using aggregate sites to estimates model parameters, but disaggregated sites to calculate welfare measures?

McConnell, K. E., "Consumer Surplus from Discrete Choice Models," Journal of Environmental Economics and Management, 29(3), Part 1 Nov. 1995, 263-70.

1. Which earlier journal article is the standard reference for how to calculate welfare effects in a discrete choice model?
2. Why is the concept of a budget constraint problematic in models that focus on single-choice occasions?
3. What intuition does McConnell use to argue that choice probabilities as a function of prices allow computation of a welfare measures equivalent to compensating variations from welfare assessment in a standard conditional logit model.
4. Does the same intuition hold for determining the welfare effects of removing several alternatives at once? For changing the attributes of alternatives?
5. Why is numerical integration used to demonstrate the equivalence of compensating variation and the proposed consumer surplus measure in the context of a nested logit model?
6. What is the conclusion with respect to whether we need worry about the awkward interpretation of the consumer's budget constraint in a single-choice occasion model?

Parsons, George R.; Kealy, Mary Jo, "A Demand Theory for Number of Trips in a Random Utility Model of Recreation," Journal of Environmental Economics and Management, 29(3), Part 1 Nov. 1995, 357-67.

1. How do the authors characterize the typical data set concerning recreational demand?
2. Bockstael, Hanemann, and Kling (BHK) used data like this. How did their model for the number of trips incorporate information from the random utility choice model across sites?
3. Is the whole story in Parsons and Kealy fully utility-theoretic, reflecting a common set of preferences across all decisions?
4. Do the site choice model and the demand model pertain to the same time frame?
5. How might the trip demand function be estimated?
6. How are changes in site attributes transmitted to the trip demand function?
7. What is the role of Parsons and Kealy's additional assumption of diminishing site utilities with increasing numbers of trips? Does this assumption affect the site choice model for any given trip?
8. Instead of passing just the inclusive value from the choice model to the demand model, what two quantities do Parsons and Kealy use? What probabilities are used in calculating these expected values?
9. What kind of utility function underlies the trip demand model? How can the first-order conditions for optimization be interpreted?
10. What is the story about numbers of trips possibly decreasing with a improvement in site characteristics? Is this result precluded in the present model?

Feather, Peter M., "Sampling and Aggregation Issues in Random Utility Model Estimation," American Journal of Agricultural Economics, 76(4), November 1994, 772-80.

1. Feather characterizes RUM models as one subclass of what other type of nonmarket valuation model?
2. Feather compares a "complete alternatives" RUM with what three alternative models?
3. What is "importance sampling"?
4. What are some reasons for aggregating alternatives?
5. In theory, what conditions are necessary for an aggregate model to be unbiased?
6. Why are three "draws" included for each randomly selected alternatives model?
7. What is the difference between the "simple" and "adjusted" models using aggregated data?
8. Which performs better, random or importance-sampling?
9. Which performs better, simple or adjusted aggregated models?
10. Is it important that the researcher's perception of an individual's choice set is not necessarily the same as the individual's perception of their choice set?

Parsons, George R.; Kealy, Mary Jo, "Randomly Drawn Opportunity Sets in a Random Utility Model of Lake Recreation," Land Economics, 68(1), February 1992, 93-106.

1. Why is the number of alternatives in a random utility choice model sometimes problematic?
2. In a conditional logit choice model, is the dimension of the parameter space affected by the number of elemental alternatives in the choice set? Why not?
3. Do packaged multiple logit or nested logit estimation algorithms have constraints on the numbers of alternatives?
4. Describe the process of sampling from alternatives.
5. The authors discuss the biases that could arise if each individual is not fully aware of all sites that are available. What type of experiment would be necessary to assess the bias resulting from including "unknown" sites in an individual's choice set?
6. What is the HYWAYS/BYWAYS software?
7. Are common preference parameters calculated for everyone? Or are preferences allowed to differ systematically by primary use mode of these sites? How?
8. The welfare estimates calculated are "equivalent variations per choice occasion per individual." Are these stable acrros the size of the randomly drawn choice sets?
9. What notable empirical finding is reported regarding values of clean water accruing to non-contact users of these sites?

Parsons, George R.; Needelman, Michael S., "Site Aggregation in a Random Utility Model of Recreation," Land Economics, 68(4), November 1992, 418-33.

1. What is the standard literature citation regarding the problem of aggregation in a Random Utility Model?
2. What two terms are commonly omitted from RUM models on aggregated sites? Why?
3. What strategy is used by these authors to assess the potential bias from omitting each of these terms?
4. How is the "coefficient of inclusive value" in a Nested Logit model pertinent to the problem of aggregation bias?
5. How is the price of an aggregate alternative calculated? How about other variables?
6. How is the performance of the aggregate models (the County Model, the Regional Model) in approximating benefits?
7. Is it possible to draw general conclusions about the bias due to aggregation of sites in a RUM model? Why or why not?
8. What four types of advice to the authors offer in their conclusions?