The roles of uncertainty and beliefs in the economy

In this dissertation, I examine the roles of uncertainty and beliefs in the economy. In chapter 2, I investigate whether macroeconomic uncertainty affects monetary-policy decisions in the US. Eight times per year, the Federal Open Market Committee (FOMC) meets to review monetary policy in light of current economic conditions. I assume that the FOMC members use a standard macroeconomic model to make sense of the economic conditions. It captures the relationships between economic growth, inflation, and the interest rate. I further assume that the policymakers are Bayesian learners: as new data comes in, they update their beliefs about the model’s parameters. These beliefs are represented by a probability distribution. I derive a measure of macroeconomic uncertainty from its dispersion. In constructing this uncertainty measure, I use macroeconomic data as it was available at each FOMC meeting. I estimate the impact of this real-time, Bayesian measure of macroeconomic uncertainty on the FOMC’s interest rate decisions. I find that policymakers set a significantly lower interest rate in times of higher macroeconomic uncertainty. In the third chapter of this dissertation, I investigate how risk attitudes influence beliefs. I adopt the heuristic switching model, in which economic agents choose between simple forecasting rules to form beliefs. I introduce a role for risk aversion in agents’ choice between these rules: they choose a rule based on its performance and the variability of that performance. Agents have different risk preferences, and therefore choose different rules, leading to heterogeneous expectations. To empirically validate the model, I draw the agents’ risk aversions from a distribution based on survey data. I incorporate this belief-formation model in a stylized financial market. I prove that a representative agent cannot capture this model. Simulations show that the resulting belief dynamics can drive unpredictable booms and busts in the asset price. Introducing small stochastic price shocks leads to larger asset price bubbles and can destabilize markets. In chapter 4, I propose an explanation for the mixed results from studies about the role of sentiment in economics: these studies measure different dimensions of sentiment that have distinct macroeconomic impacts. To test this hypothesis, I rely on Rational Beliefs theory. It implies that sentiment can be measured as the difference between observed forecasts and non-judgmental forecasts based on the available data. I use observed forecasts from the Survey of Professional Forecasters, covering 50 years, approximately 40 forecasters per survey, various economic variables (e.g., output, prices, interest rates, housing), and multiple forecasting horizons. I approximate the non-judgmental forecasts by collecting a large panel of real-time data covering all relevant aspects of the economy and using a statistical model to produce predictions. I use factor analysis to identify three dimensions that together capture about 50% of forecasters’ sentiment. I find that sentiment is indeed multidimensional, with the first dimension explaining only about a fifth of its variation. I furthermore find that each dimension has a distinct macroeconomic impact, supporting my hypothesis. My results also indicate that the survey forecasts are not always rational