Date of Graduation
Doctor of Philosophy in Environmental Dynamics (PhD)
John C. Dixon
Jennie S. Popp
Second Committee Member
Fiona M. Davidson
Environmental economics, Geographically weighted regression, Urban ecosystems, Water resources, Willingness to pay
Two decades of rapid urban growth and increasing per capita water consumption has left water providers in Northwest Arkansas concerned about their ability to meet future demand for water. Beaver Water District (BWD) is the largest of four regional water providers that draw from Beaver Lake, the only source of potable water in the region. Growth projections and per capita consumption patterns indicate that BWD could exhaust its raw water allocation as early as 2031. Municipal water customers served by BWD were surveyed about their stated priorities for water use, their water conservation behaviors, and their attitudes and perceptions about urban growth, water resources, and willingness to pay fees for future water availability. A logistic regression, or "logit" model was developed from the survey to identify the attitudes that affect willingness-to-pay (WTP) and estimate mean WTP for water. Logit models are commonly used to estimate mean WTP for an environmental good, but the parameter estimates generated in a logit model describe a global relationship between the dependent and independent variables. Because the independent variables often represent Social processes that are spatially non-stationary, geographically weighted regression (GWR) provides a means to examine spatial variations in a model by calculating local parameter estimates for every data point. GWR was applied to the data to identify spatial variations in mean WTP and the attitudes that influence WTP. Using a logit model, mean WTP was estimated at $227 annually for the region, with subsets of the data by municipality showing WTP ranges from $209 to $245. Using GWR, mean WTP over the region was estimated at $229 annually, but actual WTP values calculated for each data point ranged from $67 to $442 per year, with a clear pattern of low values correlating to closer proximity to the water resource. Both the logit and the GWR models were successful in estimating mean WTP over a region, but GWR also illustrated important spatial patterns in the data that can enrich our understanding of the problem and lead to more successful long-term strategies and solutions for meeting the future needs of a rapidly-expanding population.
Dennis, Robyn Lane, "Spatial Variations in Willingness to Pay for Water at the Local and Regional Scales Using Geographically Weighted Regression" (2011). Theses and Dissertations. 65.