Are Stated Expectations Actual Beliefs? New Evidence for the Beliefs Channel of Investment Demand

Despite growing interest in expectation surveys, critics argue that survey responses are not reliable measures of the expectations underlying financial decisions, and empirical work often finds only a weak correlation between investment and stated beliefs. In this paper, we document a systematic gap between an individual's own forecasted returns and the beliefs used in investment decisions. In particular, perceived past housing returns predict individual real estate investment decisions even conditional on flexible controls for an individual's forecasted distribution of future home-price growth. Many respondents acknowledge that they rely on past returns more than their survey-reported expected returns when making investment decisions, ruling out simple measurement-error explanations. To interpret these findings, we present evidence that investment decision-making induces cognitively uncertain investors to rely more on belief factors in which they are relatively confident, such as their estimation of recent local housing returns. Survey respondents that rely on past returns more than their surveyed forecasts frequently cite uncertainty about other belief factors as their rationale.