We study how biases in expectations vary across different settings, through a large-scale randomized experiment where participants forecast stable random processes.The experiment allows us to control the data generating process and the participants' relevant information sets, so we can cleanly measure forecast biases.We find that forecasts display significant overreaction to the most recent observation.Moreover, overreaction is especially pronounced for less persistent processes and longer forecast horizons.We also find that commonly-used expectations models do not easily account for the variation in overreaction across settings.We provide a theory of expectations formation with imperfect utilization of past information.Our model closely fits the empirical findings.
Stored only when both papers are in the corpus — 27 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.