Essays on Uncertainty and Macroeconomic Dynamics

Published in American University, 2021

Committee

Chair: Dr. Xuguang Sheng, Department of Economics, American University
Member: Dr. Gabriel Mathy, Department Economics, American University
Member: Dr. Scott Baker, Kellogg School of Management, Northwestern University
Reader: Dr. Svetlana Makarova, School of Slavonic and East European Studies, University College London

Abstract

In this dissertation I examine economic uncertainty, particularly from the perspective of disaggregation below the national level. In part one I outline the building of news-based uncertainty measures for all 50 US states plus Washington DC. I analyze different search specifications, finding that the simplest set of terms involving one group for “economy” and one group for “uncertainty” has the highest resolution, while yielding similar results to more focused news searches, including one similar to the ubiquitous policy search from Baker, Bloom and Davis (2016), and an alternative specification designed to eliminate false positives. I also explore the differences between two other national uncertainty measures - the VIX stock market volatility index from the Chicago Board Options Exchange, and the unforecastable macroeconomic factors from Jurado, Ludvigson and Ng (2015).

In part two, I build upon the analysis of the state uncertainty measures begun in part one. I show that analyzing uncertainty at the national level obscures significant state-level variation, with state-to-national uncertainty correlations ranging from 0.124 to 0.913, while some inter-state uncertainty correlations even turn negative. I then show that VAR analysis using state-level unemployment figures yields impulse response functions that are remarkably similar to existing national-level uncertainty research, with 92% of states exhibiting a rise in unemployment that peaks near 12 months after an uncertainty shock, then overshoots the starting point for a time.

Finally, in part three I attempt to get at the causal effects of an uncertainty shock, as VAR analysis is unable to do, by applying a difference-in-difference framework to the natural experiment of military base closures. I look at the 1991, 1993, 1995, and 2005 rounds of the Base Realignment and Closure (BRAC) process, whereby the government moves military jobs in and out of bases through a process that is, at least initially, heavily insulated from confounding economic indicators and political influence. This creates asymmetric and exogenous uncertainty shocks in places with different military employment, which I use to show that a one-percentage point higher military share of employment causes a tenth of a percentage point higher unemployment rate in a given Metropolitan Statistical Area (MSA) during a BRAC round.