Three essays on social dynamics and land-use change: framework, model and estimator

Simulations of future regional land-use change help planners and policymakers understand how scenarios with alternative public policy and investment choices will play out in the future, especially in conjunction with different economic and demographic trends. Models used to simulate land-use change are driven by physical, economic, infrastructure, environmental and geographical factors. The magnitude of growth is determined exogenously, independent of the existing land-use and socio-economic conditions. Even though significant relationship among income inequality, racial segregation, housing abandonment, intra-regional migration, school quality, amenities and urban growth have been established, the current generation of land-use change models does not explicitly model these social dynamics. This omission results in underestimation of new housing construction, and failure to account for urban decline, sprawl, intra-regional migration, regional imbalances and social externalities. The research described in this dissertation seeks to address this omission and is presented as three essays. The first essay explores how the spatial pattern of socio-economic characteristics determines the magnitude and location of growth and is shaped by it. It presents a framework for modeling the relationship among regional inequalities, urban distress and growth. The framework is tested using data from the St. Louis region. The second essay uses the framework to model the impact of the social dynamics on location and magnitude of growth and decline in the region. The forecasts from this model, together with population and income forecasts from a regional economic model, are used to derive probabilities of development and decline at the Census tract level and to assess the total magnitude of growth endogenously. The last essay evaluates the performance of the full information Feasible Generalized Spatial Three Stages Least Square estimator used to estimate the model presented in the second essay. Using Monte Carlo experiments, the sensitivity of results to varying degrees of spatial dependences, choice of spatial weight matrix, sample size and variance covariance matrices is analyzed. These Monte Carlo simulations provide confidence in the results of the social dynamics model