Three essays in regional economic modeling

My dissertation is about using regional economic modeling for economic impact analysis, forecasting, and for better understanding the local economy. In the first essay, I developed a nonspatial version of a community policy analysis model for Missouri counties. The model recognized the intersectoral linkages in the Missouri economy. The model consists of four modules: labor market, demography, housing market, and local public finance. Employment and total personal income drive the model. The model predicts reasonably well that increases in local employment lead to increases in local population, housing demand, local revenues, and demand for public services. However, by not considering the effects of space, the impact analysis and forecasting capabilities of my first model may not be as accurate as needed. Therefore, in my second essay, I introduce a spatial dimension into my model by specifying and estimating generalized spatial three-stage least squares procedures. The results show significant cross-county interactions within Missouri in terms of the supply of public goods, labor mobility, retail trade, and the choice of residential location. In my third essay, using South Korean regional data, I compared the forecasting accuracy of non-spatial, spatial lag, spatial error, and spatial lag and error models using in-sample data. I also compared the impact estimates of nonspatial and spatial models. The spatial components appear to improve the accuracy of the intra-county impacts. It appears that the estimated parameters tend to be sensitive to the specification of weight matrices, if the sizes of spatial units are heterogeneous and vise versa. CHAPTER I GENERAL INTRODUCTION Economic forecasting and impact studies are often part of the everyday practices of policy makers at state and local levels. When a new plant locates in a certain county, policy makers are interested in the answers to several questions. What happens to unemployment? What happens to out-commuting and in-commuting? What happens in the housing market? How will local expenditures on police, education, and fire protection be impacted by the changes? And what could happen to property taxes, sales taxes, and other sources of local revenue? Most past researches have focused on a single market or at best two (the local finance and labor market) to estimate these impacts. Decentralized governance leads to increased responsibilities on the part of local government for economic development, land use, natural resource management, education, healthcare, and public safety to name a few. As a result, local communities’ demand for decision support tools has been increased to make more informed decisions. They not only need more complex and complete analytical tools at their disposal but also need tools that can address the issue at different dimensions including spatial, temporal, distributional, and sectoral. In many instances, incorporating labor market variables and demographic variables together with local finance and housing market variables and accounting for spatial interactions among these variables gives a more accurate picture of a local economy than the analysis of a single market in isolation (Gyourko and Tracy, 1989 1991; Roback, 1982). Policy makers and economic planers have long been involved in forecasting and impact analysis efforts and analyzing their meaningfulness. Thus, a better