Essays in Macroeconomics and Entrepreneurship

This dissertation is comprised of three chapters in macroeconomics, entrepreneurship, and heterogeneous agent models. In the first chapter, I answer two main questions --- What are the empirical facts regarding entrepreneurial investment choices, and to what extent do the investment choices of entrepreneurs help to explain top wealth inequality. To that end, I document some novel facts about entrepreneurial investment dynamics. I show that these facts are suggestive that capital illiquidity are an important friction with regards to entrepreneurial investment choices. To quantify these frictions, I construct a new general-equilibrium heterogeneous agents model of entrepreneurship that features illiquid investments. I calibrate this model to identifying features of the data, and find a large role for illiquidity. I also find that the investment and savings choices of entrepreneurs help explain a substantial fraction of wealth inequality. Counterfactual analysis shows that the illiquidity friction generates substantial welfare and productivity losses by allocating wealth away from high productivity entrepreneurs to low productivity entrepreneurs, which simultaneously leads to lower wealth inequality. As such, I find that a policy of partial insurance against illiquidity risks can help ameliorate these losses, and simultaneously increases wealth inequality. In my second chapter, I present new evidence regarding the effect of uncertainty shocks on firm startup and exit rates. I document that uncertainty shocks are strongly and negatively correlated with firm startup rates, but essentially uncorrelated with exit rates. I show how my model of illiquid entrepreneurial investments can help explain these facts, and argue that capturing the extensive margin of adjustment to uncertainty shocks is important in amplifying and propagating the effects of uncertainty shocks. Finally, in my last chapter, I present a new computational algorithm to compute distributions in heterogeneous agent models. I show how this algorithm improves on current methods by reducing the amount of computational memory and time required, and provide a simple and intuitive explanation as to how this algorithm improves on the textbook method.