UC Berkeley

Does inequality react to stabilization policies and macroeconomic shocks at business cycle frequencies? Does an unanticipated innovation in inequality impact aggregate demand and drive cyclical fluctuations? Does the level of inequality influence the propagation of stabilization policies? To answer these questions, I first investigate which factors of earnings distributions are represented by a measure of inequality in Chapter 1. Chapter 2 develops an econometric tool to evaluate the contribution of macroeconomic shocks to the dynamics of an endogenous variable of interest. Finally, Chapter 3 deals with the questions above.

In Chapter 1, I derive principal components of log earnings distributions in the U.S. and propose a simple three-factor model to rationalize my empirical results. Using data on earnings distribution in the U.S. from 1978 to 2013, I find that more than 90 percent of the total variation in the distribution can be summarized by two underlying factors, which are related to the location and dispersion of the distribution, where the dispersion factor is tightly associated with the log P90/P10 index. Moreover, most of the remaining portion is due to another factor characterizing asymmetric components. To rationalize these findings, I suggest asymmetric Laplace distributions as a model of log earnings distributions. In this model, the right-tail of earnings always follows a Pareto distribution, unlike log normal distributions. Furthermore, it is a tractable distributional family featuring three parameters representing the location, dispersion, and degree of asymmetry, respectively. I describe the dynamics of those parameters in the U.S. both in trends and concerning business cycles. Finally, I illustrate how a conventional Gaussian AR(1) model for individual log earnings can be easily modified to admit asymmetric Laplace distributions.

Chapter 2 is based on joint work with Yuriy Gorodnichenko, which is forthcoming in Journal of Business and Statistics under the same title. We propose and study properties of an estimator of the forecast error variance decomposition in the local projections framework. We find for empirically relevant sample sizes that, after being bias-corrected with bootstrap, our estimator performs well in simulations. We also illustrate the workings of our estimator empirically for monetary policy and productivity shocks.

Chapter 3 deals with questions on the relationship between business cycles and earnings inequality. For an empirical investigation, I construct a novel, high-quality, quarterly measure of earnings inequality and document the following facts. First, an expansionary productivity shock and a contractionary government expenditure shock reduce earnings inequality significantly at the medium-run, while monetary policy shocks have little effects. Second, an unanticipated positive innovation in earnings inequality, which summarizes redistribution from the poor to the rich, lowers aggregate demand substantially in a U-shaped manner. Lastly, the power of stabilization policies increases with the level of inequality. To rationalize these results, I develop a tractable, theoretical framework. I analytically illustrate that inequality in a simple two-agent model is related to demand shocks in a representative agent framework. To match the shape and magnitude of the empirical impulse responses, I further introduce new features including countercyclical earnings risk, an endogenous extensive margin of being credit constrained, and decreasing relative risk aversion preferences.