I study the role of heterogeneity and idiosyncratic risk in Macroeconomics, and their implications on problems of income taxation. In the first chapter, I study the effects of redistributive taxation in an incomplete market economy with heterogeneous agents and idiosyncratic risk. I focus on the role of distortions in labor supply decisions and the interplay of heterogeneity and uninsurable idiosyncratic shocks, conducting the first general equilibrium analysis of a Negative Income Tax (NIT).I show that a NIT is a serious candidate to replace the current income tax in the United States. I find that the optimal NIT has a marginal tax rate of 28% and a transfer of 10% of per capita GDP, roughly $4600.The welfare gains of replacing the current US income tax with a NIT are equivalent to a 6.3% increase in annual consumption in every state of the world.Low-ability agents, in the bottom quintile of the productivity distribution, benefit the most, while high-ability agents are worse off. A consequence of the reform is that the composition of the labor force changes, with high-productivity agents working more, in relative terms, than low-productivity agents. Finally, I find that the riskier the economy, the higher the welfare gains of the NIT as a provider of public insurance.In the second chapter, I study labor income dynamics over the life cycle and introduce a novel methodology that can detect the presence of patterns in the idiosyncratic earnings shocks and recognize economic forces in action. Using a sample from the Panel Study of Income Dynamics (PSID), I estimate a Bayesian Logistic Smoothed Transition Autoregressive model of order 1 (LSTAR(1)) with a rich level of heterogeneity in the innovations. I find that there is a life-cycle pattern in the earning shocks: before the age 29, young workers experience shocks with higher variance and a positive probability of lower persistence than older workers. A comparison with conventional models shows that an incorrect model specification introduces bias in the estimates. The proposed model can be easily approximated with a discrete Markov process. This means that this model can be used by macroeconomists to calibrate income processes. Abstract Approved: Thesis Supervisor To my loving parents and dear family ii All theories are valid and none is relevant. What makes them relevant is what you do with them. -Jorge Luis Borges, Complete Works iii ACKNOWLEDGEMENTS An exhaustive list of all of the people who contributed to this dissertation will never be complete, no matter how hard I try. There are many friends and professors to whom I owe a great deal of gratitude. I will start with Gustavo Ventura, my main advisor and first macroeconomics professor at Iowa. He taught and pushed me to be a better economist, always curious and in search of good questions, fearless of the difficulty of any economic problem.He has been a major role model for me, and an outstanding mentor.Charles Whiteman has also been an inspiration for me. He spent time with me, correcting my thinking, encourag...