| Nearly all developed countries have income tax schedules that are progressive at the margin. Despite this fact, models of flat income taxation dominate the growth literature so that the effects of progressive taxation on savings are not well-understood. In the three essays of this dissertation, I find that abandoning the assumption of flat taxation leads to new results regarding the shape of the long-run distributions of income and wealth, the support for redistribution, and the nature of business cycle fluctuations.;The first essay considers the long-run distribution of capital holdings in a model with complete asset markets and progressive taxation. With homogeneous discount factors, heterogeneous labor productivity, and monotone-increasing marginal tax rates, all non-constrained households must have the lowest income in the population. This necessitates that capital and labor income are negatively correlated as are income and wealth if assets have a lower bound that binds in equilibrium. This paper concludes by showing that the introduction of heterogeneous discount factors can make the model's predictions more consistent with data.;The second essay examines the degree of income tax progressivity chosen through a majority vote in the neoclassical growth model. Households differ with respect to their labor productivity and their subjective discount factor. Unlike previous studies, this model endogenizes both the tax policy and the income distribution. A simulation of the model shows that support for a high degree of income tax progressivity is greater than would be suggested by static models. In the long run, increased progressivity effects only a small decrease in income and wealth inequality.;The final essay studies sunspot fluctuations in a model with heterogeneous households. Wealth inequality reduces the degree of increasing returns needed to produce indeterminacy, while wage inequality increases it. When the model is calibrated to match the joint distribution of hours, income, and wealth, the required degree of increasing returns to scale is still much too high to be supported empirically (although smaller than similar homogeneous agent economies). The model robustly predicts only one sunspot, despite having 1262 predetermined state variables. |
