Estimating the demand for and value of recreation access to national forest wilderness: a comparison of travel cost and onsite cost day models

by Jianping Zhu

Institution: University of Georgia
Department: Statistics
Degree: MS
Year: 2007
Keywords: Wilderness Recreation
Record ID: 1811172
Full text PDF: http://purl.galileo.usg.edu/uga_etd/zhu_jianping_200705_ms


The Travel Cost Method (TCM) is the dominant method for recreation demand analysis. The number of trips to a site is assumed to relate to travel cost, travel time and other demographics. One disadvantage of the conventional TCM is that it does not recognize potential spatial limitations. Bell and Leeworthy (1990) developed an alternative model to deal with the spatial problems associated with visitors coming from significant distances to use principally beach resources. Instead of modeling annual trip numbers to the recreational site, the number of days spent at the site is modeled. This model is called on-site cost model (OSCM ) or expenditure day model. In both TCM and OSCM, the response variables are count data based on an on-site survey. In this thesis, the OSCM, introduced by Bell and Leeworthy (1990), is applied to visitation at National Forest Wilderness areas. In both cases, the estimators need to account for the fact that the dependent variables are non-negative integers under truncation, overdispersion, and endogenous stratification. Additionally, the more conventional TCM is applied to these same data. Results from both models are compared and the models are tested with respect to derived welfare estimates. Visitors’ total Wilderness days are modeled as a dependent variable on several demographic and socioeconomic factors. This model includes visitors who are from a significant distance and those from a short distance. A statistical justification about which assumptions are violated is provided by applying Ordinary Least Square (OLS) regression model to recreation count data models. Then instead of OLS, negative binomial regression is applied to build the models. Structured weights were incorporated into the models to reflect the nature endogenous stratification. The consumer surplus estimates for different models are derived by using the bootstrap method, and statistical tests are used to examine their congruency. The findings of this research in general suggest that there is no significant difference in net economic value on per individual per day basis between TCM and OSCM.