Behavioral biases, information frictions and interest rate expectations

Abstract

We use expectations of the short rate inferred from the term structure of interest rates to test several well-known models of behavioral biases and information frictions. We classify signals about future short rates by their cost of acquisition and find evidence of overreaction to high-cost signals and underreaction to low-cost signals, providing support for the overconfidence bias. We show that our results are unlikely to be driven by time-varying risk premia. The biases are so large that the market’s forecast errors are larger at all horizons than for forecasts obtained by assuming that the short rate follows a random walk.