Geospatial Economics

Abstract

Understanding how economic systems interact with ecosystems requires models that include geospatial heterogeneity. Integration of economic and ecological systems is vital for sustainable development, especially with respect to climate change, food security and the management of common property resources. Advances in remote sensing and geographic information systems have created a wealth of data applicable to economics, but it is challenging to incorporate high-resolution, global data in existing economic models. In this thesis, I integrate geospatial data with economic theory to analyze important environmental problems. The next three chapters describe the techniques I use for modeling geospatially-explicit economic systems and apply them to current environmental challenges. Chapter one addresses the tradeoffs between food production and environmental protection. I address the question of how we can optimally feed a growing population (requiring a 100% increase in calorie production by 2050) while minimizing the loss of carbon storage (which is important for mitigating climate change). I use high-resolution, gridded global data to give geospatial specificity to the results of the optimization. The framework I present in this chapter includes only one production choice and one environmental good, but it is more broadly applicable to multiple goods and multiple ecosystem services. This chapter also shows how using geospatial data can increase the policy relevance of an analysis. For example, instead of claiming that tropical forests are in general very valuable, using geospatial data allows for the more precise claim that this specific 10 kilometer patch of forest is better kept as forest than cultivated. Spatially explicit information like this can help construct more specific policies, such as food-for-carbon swaps or identifying which parcels ought to be protected first given a limited conservation budget. Chapter two presents a microeconomic model of spatial foraging that addresses how humans gather goods on a spatially heterogeneous landscape when transport costs are non-negligible. The general problem of foraging arises when multiple agents located in space compete for resources that are characterized by their location. I use agent-based simulation methods to account for agents that must move over the landscape subject to terrain and road networks, depletion by rival agents and spatial heterogeneity with a large number of agents (10 million). The model is applicable to several topics in environmental economics, including fuelwood collection and fisheries management, but also more general economic topics such as housing, employment search, transportation, pollution and urban economics. Another contribution of chapter two is that it makes several methodological advances that allow for spatially-explicit agent-based simulation on extremely large systems. These advances are of two types: first, I present data creation methods that allow for high resolution data to be created globally, relying on satellite-derived data products and spatial downscaling techniques to estimate environmental and social indicators, such as population density or spatially defined wages. Second, I identify computational methods (and implement them in a software application) that allows for fast calculation of agent interactions and movement of space. I discuss the data storage types necessary for this along with a method of vectorizing the calculations to enable computation of extremely large systems (with as many as 10 billion agents). Chapter two concludes an application of the spatial foraging model in which I assess how villagers in Tanzania gather firewood from forests. Firewood collection is a useful example because the need for a spatially explicit model is clear. Transportation costs of firewood are very high relative to their value and firewood in almost always collected by agents foraging for their own consumption. For instance, Fabe and Grote (2013) report that 97.5% of household in Tanzania use firewood as their main fuel, but only 13% of firewood is purchased. I define behavior rules in the simulation based on a microeconomic agricultural household production model (Singh et al. 1986, Bardhan and Udry 1999) apply them to high-resolution geospatial data for Tanzania. I iteratively simulate individual agents' foraging actions and observe the value of firewood obtained. The estimates obtained from this method match existing estimates of firewood collection while providing more detail about where the firewood is collected. The empirical application presented in chapter two is useful to practitioners of ecosystem serve estimation. Calculating the ecosystem service value of non-timber forest products (NTFPs) has been difficult in practice due to computational and theoretical problems. Existing estimation approaches identify the abundance of NTFPs on different forest types but do not explicitly state how agents gather the firewood. The approach I use accounts for how these gathering decisions interact with the abundance of NTFPs to determine the ecosystem service value. The final chapter in this thesis presents a theoretical model that that incorporates reciprocity in a utility maximization model to analyze common-property resource dilemmas. This model, which I refer to as the commons reciprocity utility model, allows agents to make interpersonal comparisons of utility in order to reward cooperators and punish detractors. Extending traditional utility theory in this way is useful to describe the wide-spread observation that individual economic agents do not always free-ride and do not always fall for the tragedy of the commons (Hardin 1968; Ostrom 1990). Although this chapter is primarily focused on theory, I provide two environmental examples to illustrate how it can be applied (including full details in Appendix 2). First, I discuss how international negotiations on climate change can be modeled in this framework by describing each nations' decisions to meet their emissions abatement targets in reciprocal terms. Specifically, nations will reward other nations who do meet their abatement goals and will punish those who do not. I provide a numerical example of this situation that shows increased levels of abatement, higher than the prediction of strong free-riding. Second, I apply the commons reciprocity utility model to a forest commons to explore how agents' decisions to engage in sustainable forestry or to clear-cut the forest depends on the reciprocal relationships of nearby agents. This example is preliminary, but shows how the model can be applied to the agent-based simulation techniques introduced in chapter two. At the deepest level, the goal of this thesis is not to present a complex system of models, but is to answer the question of "how ought we live?" As humans harness an ever-greater portion of available energy and focus it into ever-more complex arrangements, the question of understanding our place in our environment grows more challenging. It requires modeling economic behavior in conjunction with the geospatial landscape on which we act. It is my hope that the methods presented here help condense the nebulous connections among economic and ecological systems into useful bits of truth.