A Nonparametric Test of the Traditional Travel Cost Model

  • Author(s) / Creator(s)
  • The travel cost model of recreation demand has been used to study the services provided by recreation sites for over 30 years. Numerous assessments have been made of alternative ways of implementing the travel cost model (TCM) and of its performance according to a variety of criteria. The vast majority of these assessments are \"parametric\" in nature, i.e., they are based at least in part on specific assumptions regarding functional forms and error distributions in demand equations. Recently, Adamowicz and Graham-Tomasi proposed a \"nonparametric\" approach to testing the TCM based on an examination of the consistency of recreation data and the TCM with revealed preference axioms of rational choice. The nonparametric approach has the advantage of expurgating from the test procedure potentially confounding parametric assumptions. In an empirical application of the methods they propose, Adamowicz and Graham-Tomasi (AG-T) found that recreation data on prices and quantities for Bighorn sheep hunting in Canada, when combined with a particular form of the TCM, generally were consistent with the axioms of rational choice. Very few violations of revealed preference axioms were identified, and when they did exist, a measure of the severity of the violation showed them to be \"small.\" The version of the TCM they tested involved a separability assumption across trips, so that different trips were taken to be independent choices. While this separability assumption often is invoked in travel cost analyses based on the \"discrete choice\" model (e.g. Bockstael et al. 1987a), it is not imposed in the traditional TCM, in which the choice is the number of visits over a recreation season. In this paper we provide a nonparametric test of the traditional TCM. We examine a multi-site model with choices made regarding the number of trips to take to each site over an entire recreation season. We employ the data on Bighorn sheep hunting used in AG-T. However, we find much weaker consistency of these data with the axioms of choice than did AG-T; here, violations are numerous and \"large.\" Subject to some qualifications which we discuss below, this result calls into question the correspondence between actual recreation choices and their representation in the version of the traditional TCM we employ.

  • Date created
  • Subjects / Keywords
  • Type of Item
  • DOI
  • License
    Attribution-NonCommercial-NoDerivatives 3.0 International