How Do Population Growth, Land-use Regulations, and Precipitation Patterns Affect Water Use? A Fine-scale Empirical Analysis of Landscape Change

Graduation date: 2015A growing body of literature exists on how human population growth and changes in climatic factors influence the availability of water (Elliot et al. 2014, Prudhomme et al. 2014). These studies typically conclude that climate change is expected to have negative consequences on water availability, an effect that is magnified or exceeded by continued growth in human populations (Vorosmarty et al. 2000, Mcdonald et al. 2010). However, there are several deficiencies associated with how these existing studies from the natural science literature analyze the interconnectedness of the human-water-climate system. For one, the scale of the models that are employed often precludes researchers from accounting for the location-specific nature of how water is used. Failing to recognize the inherently spatial nature of water use compounds the inability of researchers to analyze the role of other spatially-dependent behavioral processes that are associated with population growth, and hence water use. The same can be said for the treatment of climate change in the existing literature. If the effects of climate change on water use are transmitted through intermediate human behavioral processes, ignoring the indirect nature of the relationship will not accurately convey the true localized outcomes that may be realized. Moreover, there may be policies in place that govern how certain spatial processes operate, which again cannot be accounted for without recognition of the fine-scale spatial aspects of water use.\ud In this dissertation, I analyze the effects of population growth, land-use regulations, and climate change on localized consumptive water use through the lens of land-use change, an inherently spatial process. Intuitively, as a city's population expands and more land is developed at its extensive margin, the total amount of water consumed by the residential sector will correspondingly increase. However, I show that the net effect of land development on the total amount of water used by the agricultural and residential sectors combined depends primarily on three factors: (1) the amount and location of rainfed versus irrigated agricultural land within the vicinity of the city, (2) the rate of population growth occurring within the city, and (3) the spatial pattern associated with where new land development occurs. Somewhat counterintuitively, what I find is that under certain conditions more sprawl-like development patterns are associated with lower total water use. This result stems from the fact that low-density development patterns tend to be associated with increased conversion of irrigated agricultural land. Given the fact that irrigated agricultural land requires relatively more intensive water use when compared with urban land, lower density urbanization patterns can result in a reduction in total water use if the city expands into lands that are predominantly used for irrigated agriculture.\ud My empirical approach combines fine-scale econometric and simulation methods in order to accommodate the spatial heterogeneity associated with land and water use in my study area, Oregon's Willamette Valley. In Chapter 3, I estimate a series of parcellevel\ud hedonic property value models for land in developed, agricultural, and forest uses using a novel panel data set on land values for our study area, Oregon's Willamette Valley. Although the hedonic models are not an end in themselves, several policy applications are examined vis-a-vis the estimated property value relationships for each land use. First, for developed lands, I explore the effects of Oregon's comprehensive land-use plan, with specific focus on urban growth boundary designation, through the use of a variety of panel data estimators. Second, I analyze the relationship between agricultural land values and climatic influences, with particular emphasis on the interactions between the holding of irrigation rights and the effect of growing-season precipitation. The third application of the hedonic models pertains to how forest land values are influenced by transportation costs, represented by distance to the nearest processing mill, which change over time due to the large number of mill closures during my study period, partly in response to the old-growth harvest restrictions imposed by the Northwest Forest Plan.\ud In Chapter 4, the estimated hedonic relationships are used to predict the values associated with land in selected and unselected uses in my land-use change data set, which comes from the US Geological Survey's Land Cover Trends (LCT) project. In a similar fashion to Bockstael (1996), I then use the land value predictions as explanatory variables in a set of discrete-choice urbanization models. In generating the predicted land values with the estimated hedonic relationships, I confront the sample selection issue inherent in applying the developed land hedonic model to lands that start in undeveloped uses using econometric methods suitable for panel data. I also compare the results generated with the coupled hedonic-urbanization approach with those from a reducedform model, which is more commonly found throughout the modern land economics literature. Additionally, using a Monte Carlo analysis, I illustrate the efficiency of the two-stage approach, and demonstrate how this property depends of the frequency of observed land development decisions.\ud With my fully estimated econometric framework, I construct a spatially-explicit landscape simulation model in Chapter 5, which is used to analyze total water use over the period 2000-2070. The landscape simulations feature exogenous growth in population and income as the predominant drivers of urban land value dynamics. Agricultural water use is determined by spatial data on the location of irrigation water rights administered through the prior appropriation doctrine. Citywide urban demands are determined by the size of the city, demographic characteristics, water prices, pricing structure, and the demand for outdoor water use, which is influenced by precipitation. To generate my main results, I analyze three sets of scenarios in which I alter: (1) the area upon which land development is possible and (2) the city-specific population growth trajectory, and (3) the opportunity cost of land development for agricultural landowners. The first scenario set relates to the stringency of land-use regulations, which directly affects both the density of population growth and the urban composition of the city. Population growth, which is featured in the second set of scenarios, has several competing influences on total water use, among which are a direct influence on citywide residential water demand, and also the indirect effects that come about through the incentives to develop individual land parcels. In my third set of scenarios, I hold the land-use regulatory environment constant and impose permanent changes in precipitation patterns. Growing season precipitation\ud indirectly influences total water use through its direct effect on the opportunity cost of land development for the owners of rainfed agricultural land.\ud A novel feature of the simulation framework is that population density, which enters the developed land hedonic function, is treated as endogenous, as it is a function of development patterns and the permitted levels of development under different urban containment policies. Additionally, the simulation design is stochastic in that I treat land development decisions in a probabilistic fashion, facilitating an analysis of the distribution of results, including my main results for total water use, in each of the simulated scenarios. My results indicate that under situations that stimulate increased land development, the distribution of total water use is characterized by a lower mean if there is a sufficient amount of irrigated agricultural land in the vicinity of the city boundary, and if population growth is not prohibitively high. Further, the variance of total water use is decreasing in population density, as more sprawl-like development outcomes have a relatively wider array of potential spatial patterns associated with new land development.\ud In the broadest sense, my dissertation highlights the interplay between population growth, climate change, land-use regulations, and the spatial patterns associated with both non-urban water endowments and new land development. The intermediate link that land-use change provides between climate change and population growth and water use has yet to be acknowledged and given proper treatment in the existing literature. Moreover, the connection between land-use change and water use emphasizes the inherent spatial heterogeneity in how water is used. Given the ubiquity of policies worldwide that impose location-specific constraints on water use, my analysis suggests\ud that spatial considerations should be taken into account in subsequent studies that aim to analyze localized patterns of water use and how they may evolve in the future