In my thesis I construct a four-region, four-product, three-factor, two income class applied general equilibrium model of the United States, which I use to examine several issues in sub-national tax incidence. Each of the four regions is an aggregate of several states. Within each region four products are produced; one is interregionally traded, one is not traded, one is provided by the region's government, and the last is provided by the national government. Each of the four products requires capital, skilled labor, and unskilled labor for its production. These factors are supplied by two types of households, Rich and Poor, who differ in their relative factor endowments and in their ability to relocate among regions (Rich may be mobile; Poor are immobile). Each household consumes each of the four products. Federal and regional governments tax households, firms, and products.
A benchmark equilibrium is established in which product and factor markets clear and in which Rich households have no incentive to relocate among regions. I then simulate several tax policy changes. In the first group of simulations, a single region's tax rates are (separately and unilaterally) increased, with the revenue used to fund additional government spending. One conclusion drawn from these simulations is that household utilities fall in any region which unilaterally increases any of its tax instruments. Interestingly, however, the business capital tax is the least "painful" way for a regional government to unilaterally increase its tax revenue. In the second group of simulations, a regional government unilaterally eliminates its household taxes, replacing the lost revenue by increasing its business capital taxes. In these simulations, household utilities generally increase in the region that eliminates its household taxes, while utilities decrease in other regions. This result is especially robust for regions which are net capital importers--those regions in which producers use more capital than their regions' residents own. Sensitivity analysis reveals that my simulation results are fairly robust, even when important parameter estimates (such as elasticities of substitution in production and consumption) are varied.