Francesca Bastianello (Booth): Mental Models and Financial Forecasts
Abstract: We uncover the mental models financial professionals use to explain their quantitative forecasts, and show how they shape beliefs and return predictability. Using the near-universe of 2.1 million equity analyst reports, we collect the valuation methods analysts adopt to compute their price targets, together with their reasoning, measured as attention to topics, and their associated valuation channels, time horizons, and sentiments. To validate the reliability of our output, we introduce a multi-step LLM prompting strategy and new diagnostic tools. Consistent with a model of top-down and bottom-up attention, we then uncover three sets of facts. First, analysts’ mental models are sparse and rigid, and the choice of attention allocation and valuation methods are jointly determined by both analyst- and firm-characteristics. Second, analysts’ reasoning translates into their quantitative forecasts. Both attention and valuation methods contribute to differences in valuations over time and across analysts, but variation in attention plays a bigger role. Third, we study the extent to which different topics contribute to over and underreaction to information, and show how biases in analysts’ reasoning are reflected in asset prices. Analysts underreact to macroeconomic topics, and overreact to firm-related topics, and this contributes to return predictability.
Economics
(847) 491-8200
Email