Essays on Information Frictions and Macroeconomic Uncertainty

Abstract

This dissertation studies the role of information frictions and uncertainty in the macroeconomy. The three chapters expand a basic environment with information frictions to feature (i) variable information quality, (ii) tail risk, and (iii) social learning. The first chapter introduces information quality in a real business cycle model with costly information. Using data from the Survey of Professional Forecasters, I document that forecast errors are larger during downturns, even though agents acquire more information. I then augment a rational inattention model with variable information quality. Information quality depends on both data abundance and search intensity in the demand for information. Unlike rational inattention models, in which there is perfect supply of information, I allow for time-varying data abundance, or information supply. Procyclical supply of information generates pro-cyclical information quality, which in turn rationalizes the puzzling evidence that information acquisition and uncertainty both increase in downturns. A Bayesian estimation of the model for the U.S. economy shows that information quality accounts for sizable fluctuations in uncertainty and aggregate output. The model also generates: (i) systematic biases, if agents do not take into account fluctuations in information quality, (ii) variation in information processing costs, which produces dispersion in downturns, and (iii) production externalities, as firms do not internalize that more activity generates data abundance, which reduces uncertainty. The second chapter (co-authored with Yeow Hwee Chua) studies expectations formation in a Bayesian learning framework when the environment features tail risk. First, in the presence of tail risk, second moment shocks lead to more pessimistic forecasts compared with the forecasts generated in an environment that has zero probability of tail risk events. Second, individuals overreact to first moment shocks compared to the environment without tail risk, as they reassess the probability of finding themselves in a disaster state. Third, the magnitude of the overreaction depends on the level of uncertainty. We document these theoretical predictions for the U.S. economy over the 1978 to 2016 period, using data on expectations, uncertainty, sentiment, and tail risk. The theoretical framework predicts that not accounting for tail risk would lead the econometrician to conclude expectations are biased even though agents are in fact fully Bayesian. The third chapter studies uncertainty contagion across countries in a model of endogenous information acquisition with social learning. Motivated by evidence that macroeconomic uncertainty tends to comove across countries, this chapter builds a two-country model of rationally inattentive decision-making in which each country learns about national and international conditions independently and from each other. When fundamental uncertainty exogenously rises in one country, economic agents in that country allocate more attention to local conditions and less attention to international conditions. Global uncertainty rises as a result, and agents in the other country endogenously reallocate more attention to global conditions. This results in an increase in uncertainty about idiosyncratic conditions in the other country, even though there has been no change in fundamental uncertainty in that country. The model also predicts higher uncertainty contagion during crises.