Essays on Imperfect Information in Macroeconomics

Abstract

This dissertation contains three essays addressing issues pertaining to macroeconomic policies in presence of imperfect and heterogeneous information. Chapter 1 studies how central banks should design communication as a function of the economic fundamentals and the private sector's heterogeneous beliefs about these fundamentals. Chapter 2 examines how the Federal Open Market Committee's state-dependent topics coverage may affect expectations about future monetary policy. Chapter 3 measures the impact of uncertainty about fiscal policy on financial markets. Macroeconomic decisions involve expectations about the state of the economy and the private sector relies on information provided by central banks to form these expectations. Central banks therefore have a central role in shaping these expectations. Chapter 1 presents a model in which a central bank has incentives to use communication strategically to shape expectations so that the private sector takes a specific action regardless of the fundamentals. In this chapter, I formalize these strategic motives to communicate differently across states in a Bayesian persuasion game with heterogeneous receivers. A Sender communicates about a binary fundamental to Receivers, who holds heterogeneous beliefs about the state. The Sender wants them to take a specific action regardless of the fundamental but Receivers want to align their action with the fundamental. I derive the Sender's optimal disclosure strategy about the fundamentals as a function of both the fundamentals and the Receivers' disagreement. Then, I apply this framework to a central bank communication problem and test empirically the predictions in the model using one example of communication, the Fed's forecasts. I show that a central bank would want to send moderating signals (reporting the fundamental in either state with positive error probabilities), but the reporting accuracy increases with private sector disagreement. The second chapter analyzes the extent of state-dependent coverage of topics by the FOMC. A topic's prevalence could affect expectations in two ways: first, it provides information about the fundamental but the prevalence can also provide information about how extreme the realization is. I first document, applying computational linguistics methods to FOMC minutes, that a topics' newsworthiness varies over time and depends on both variation and level of its related macroeconomic variables: negative outcomes such as high inflation, low output, high unemployment make their associated topics more newsworthy. This suggests that the minutes are potentially an informative source about what the central bank is concerned about, and thus likely to react to. I then develop a model in which this state dependent composition (unusual number of signals about a fundamental) impacts agents' expectations about both the state of the economy and the interest rate, therefore generating a signaling effect about an interest rate change. Taking into account this signaling effect of the mix of topics, I aim to derive the optimal state contingent communication policy. The third chapter, co-authored with Yu She, turns to uncertainty and how disagreement or uncertain communication from policy makers can impact financial markets. We investigate the impact of uncertainty about fiscal policy on nominal yields, such as the fiscal cliff episode of 2012 and government shutdown of 2013. Both episodes were marked by an intense debate on Twitter between politicians. We gather tweets from politicians and government agencies during the period January 2012 to December 2015 which are related to a potential shutdown. We use sentiment analysis such as dictionary methods to measure uncertainty and negative sentiment to create a proxy for government policy uncertainty. Regressing this proxy and dummies for FOMC meetings on nominal yields at daily frequency, we find that an increase in disagreement or uncertainty portrayed through the tweets has a negative impact on nominal yields (3-month to 1-year maturity).