Measuring Biases in Expectation Formation

We develop a general framework for measuring biases in expectation formation. The method is based on the insight that biases can be inferred from the response of forecast errors to past news. Empirically, biases are measured by flexibly estimating the impulse response function of forecast errors. The framework does not require precise knowledge of the true data-generating process, and it nests all major existing models of expectations. Monte Carlo simulations show that the method is able to detect biases in empirically relevant settings. We illustrate the methodology using data on inflation forecasts. Our framework can guide future models of expectations.