Macroeconomic Expectations and Limited Awareness

Abstract

We use data on individual expectations from the Survey of Professional Forecasters and Blue Chips, and document a few novel empirical facts that cannot be reconciled with the Rational Expectations hypothesis. We argue that a simple model where agents are unaware of (at least some) supply shocks and partly confound them with demand shocks can rationalize the empirical evidence. In particular, since agents are subject to an omitted-variable bias when inferring the trade-off between output and inflation, agents perceive a ``flatter'' Phillips curve—a property consistent with our evidence. Moreover, the model naturally generates a negative (unconditional) correlation between expectation revisions and inflation forecast errors—a phenomenon commonly referred to as ``expectation overreaction''. We introduce limited-awareness (LA) agents into an otherwise standard New-Keynesian model and show that, relative to the rational expectations benchmark, the model generates more volatility in output and less volatility in inflation, regardless of the type of shock. Relatedly, we also show that the presence of LA agents increases the output cost of reducing inflation. Since LA agents try to rationalize the observed movements in inflation and output within their mis-specified model, an aggregate supply shock may exhibit qualitative features that resemble those of a demand shock (even in the absence of cross-sector complementarities). Finally, we estimate a VAR that incorporates both actual and expectations data. Using sign restrictions on aggregate variables and expectations based on the theoretical predictions of the model, we identify misinterpreted supply shocks and show that they explain 20% of total GDP volatility. (Joint with Davide Debortoli (UPF), Luigi Iovino (Bocconi), and Donghai Zhang (NUS and Bonn)