Essays on recreation demand

This dissertation extends discrete-choice models to allow for the choice of multiple goods and the choice of quantities of each good as joint decisions. Traditional models of discrete choice assume that the quantity purchased of each good is constant across individuals in a population. They also assume that an individual buys only one of the goods offered. We relax both of these assumptions.;In an application to demand for recreational sites in Costa Rica, we model the choice of sites to visit as well as the length of stay. In Chapter 1, we focus on Costa Ricans. This population typically visits a single site per trip. Using a fully discrete model, we define the alternatives in the choice set so that each alternative represents a site to visit and a length of stay. In chapters 2 and 3 we focus on foreigners and allow for multiple sites to be visited during a single trip. We estimate a discrete choice model in which each alternative in the individual's choice set is a set of sites to visit for a particular length of time. By using a stated preference database, we are able to deal with the huge universal choice set that results from defining each alternative in the way described. Chapter 3 sheds light on the way in which the "No Purchase" option should be interpreted in stated preference datasets.;We find that utility from visiting a site depends on length of stay and that the marginal utility of time and marginal consumer surplus schedules vary significantly across site types. The implication is that on-site-time costs as well as individual time constraints, can affect an individual's choice of site to visit and should therefore be included in site choice models.;We also find that beaches and forests enhance each other's value when visited over the same trip, and so do rivers and volcanoes. The model also predicts a high WTP for site variety over a single trip. Finally, the demand for beaches presents the highest (positive) elasticity with respect to changes in lodging prices at forests and vice versa. Similarly, the demand for volcanoes is the most elastic with respect to lodging prices at rivers and vice versa.;We show that the model has a better fit if one allows the "No Purchase" option to be interpreted as the possibility of taking trips different from those presented in the survey. WTP estimates for the set of trips offered in the survey are higher under this proposed interpretation than under the traditional one. However, different types of sites are closer substitutes for each other and the WTP for each site type is lower under this new interpretation.