Under- and Over-Reaction in Yield Curve Expectations
I study how professional forecasts of interest rates across maturities respond to new information. I confirm and provide additional evidence for a pattern noted by Bordalo et al. (2019b): forecasts for short-term rates underreact to new information while forecasts for long-term rates overreact. I propose a new explanation based on ``autocorrelation averaging,” whereby, due to limited cognitive processing capacity, forecasters’ estimate of the autocorrelation of a given process is biased toward the average autocorrelation of all the processes they observe. Consistent with this view, I show that forecasters over-estimate the autocorrelation of the less persistent term premium component of interest rates and under-estimate the autocorrelation of the more persistent short rate component. A calibrated model quantitatively matches the documented pattern of misreaction. Finally, I explore the pattern’s implication for asset prices. I show that an overreaction-motivated predictor, the realized forecast error for the 10-year Treasury yield, robustly predicts excess bond returns.