Essays on inflation expectations and information frictions

This dissertation empirically investigates the expectations formation process and the constraints that economic agents face in forming beliefs about macroeconomic variables. Chapter 1 contributes to and extends our current understanding of information frictions in expectations. I first propose a new framework for estimating noisy information using individual forecasts, rather than mean forecasts as commonly done in previous work. This approach provides more power for identifying underlying information rigidities. I further extend this framework to incorporate misperceptions on the part of economic agents about the persistence of the underlying process being forecasted. Applying this framework to the U.S. inflation forecasts of professional forecasters points toward significantly less noisy information than previous estimates suggest but reveals a systematic underestimation on the part of forecasters of the persistence of inflation. Using a structural model that incorporates both noisy signals and misperceptions of persistence, I quantify the relative importance of each channel in accounting for the expectations formation process of these agents. The results indicate that, even for professional forecasters, there are multiple forces that generate economically significant deviations from full information. Chapter 2 is joint work with Olivier Coibion, Yuriy Gorodnichenko, and Saten Kumar. Using novel survey questions on the higher-order expectations of firm managers, we study the formation and evolution of these beliefs. A unique experimental approach allows us to characterize the degree of higher-order thinking of economic agents and how this degree of higher-order thinking affects managers' expectations as well as their economic decisions. We then relate these results to macroeconomic models in which higher order thinking matters for dynamics. Chapter 3 is develops a method for measuring the information flow of economic agents at a given point in time using survey data. I document a reduction in attention to several macroeconomic variables over time. I further document that in periods in which agents are paying more attention to a specific variable, there is also greater cross-sectional dispersion in attention across agents.