Essays on Optimal Taxation and Dynamic Mechanism Design

This dissertation is composed of three essays. In the first essay, I study optimal taxation of family income. The model considers families as composed by two income earners, cooperatively taking labor supply decisions by maximizing a unique utility function. Since productivities are private information, the model is an instance of multidimensional screening. For zero cross elasticity of labor supply, I show the full solution in terms of exogenous parameters. Properties of the solution include the classical no distortion at the top, and positivity of marginal taxes. For non zero cross elasticity, I show through an example that some of these properties might be overturned. High productivity types can be distorted and marginal taxes can be negative.;In the second essay, I study a dynamic model in which a principal and several agents interact over many periods. Agents have private information and the principal takes a decision in every period that influences agents' surplus. I study the possibility of full surplus extraction under the assumption that the principal chooses the mechanism period by period. I show that correlation of information across players is sufficient to obtain full surplus extraction, when the implementation criterion is perfect Bayesian equilibrium.;In the third essay, I return to optimal taxation and in a simple unidimensional model I study the bunching properties of the optimal tax schedule. I show that the region of bunching is non monotone in the redistributive tastes of the social planner. It can be empty for a social planner close to an unweighted utilitarian and for a Rawlsian one, while for intermediate redistributive tastes some bunching appears.