Simple Models and Biased Forecasts

TL;DR

This paper introduces a framework where agents use simple, limited-state models for forecasting, which explains observed economic biases and improves the fit of models to real-world data.

Contribution

It develops a new neoclassical synthesis model showing that simple models better match data and resolve key puzzles in macroeconomic theory.

Findings

Simple models increase decision persistence and co-movement.

Using simple models improves data fit over rational expectations.

The approach resolves the Barro-King and forward guidance puzzles.

Abstract

This paper proposes a framework in which agents are constrained to use simple models to forecast economic variables and characterizes the resulting biases. It considers agents who can only entertain state-space models with no more than d states, where d measures the intertemporal complexity of a model. Agents are boundedly rational in that they can only consider models that are too simple to capture the true process, yet they use the best model among those considered. Using simple models adds persistence to forward-looking decisions and increases the comovement among them. This mechanism narrows the gap between business-cycle theory and data. In a new neoclassical synthesis model, the assumption that agents use simple models fits the data much better than the rational-expectations hypothesis. Moreover, simple models simultaneously resolve the Barro-King and forward guidance puzzles while improving the propagation of TFP shocks.