Working Papers
Mental Models and Financial Forecasts
(with Paul Décaire and Marius Guenzel). July 2025.
Abstract. We uncover financial professionals’ mental models—the reasoning they use to explain their quantitative forecasts. We organize our analysis around a framework of top-down and bottom-up attention, where analysts endogenously choose both a valuation method and how to allocate attention across variables. Using the near-universe of 2.1 million equity analyst reports, we collect the valuation methods analysts adopt to compute their price targets. To measure attention, we then prompt large language models (LLMs) on a subset of over 300,000 reports to extract 11.8 million lines of reasoning—each combining a topic, valuation channel, time horizon, and sentiment. To validate the reliability of our output, we introduce a multi-step LLM prompting strategy and new diagnostic tools. We document five main findings. (1) Analysts exhibit sparse and rigid mental models. (2) The choice of valuation methods and attention are closely linked. (3) Attention allocation across variables plays a bigger role than valuation methods in explaining both changes in valuations over time and disagreement across analysts. (4) Biases in analysts’ forecasts are driven by over-reaction to firm-specific features and under-reaction to macro-related ones. (5) These biases translate into asset pricing patterns: topics analysts overreact (underreact) to predict lower (higher) realized returns.
Biases in Belief Updating Within and Across Domains
(with Alex Imas). August 2025.
Abstract. We study variation in over and underreaction both within and across different domains. We propose a model where bounds on attention and information processing lead people to adopt potentially distorted mental representations of their information environment. These mental representations consist of experience-based priors (which generate a default frame people approach a given problem with), and attention weights across features (which determine how much people adjust their default frame to the current information environment). Applying our framework to the canonical inference and forecasting learning domains, we show that variation in over and underreaction across domains is largely due to different contexts cueing different default frames. Variation in over and underreaction within domains is due to insensitivity to features, and an insufficient adjustment of their default frames to the underlying information environment. When varying a single feature, we recover the result of overreaction to weaker signals and underreaction to stronger signals. However, once we allow for attention to interact with multiple features, this comparative static is modulated by attention, and can even be fully reversed. When this is the case, neglect of one feature generates excess sensitivity with respect to another. Empirical tests further identify the mechanism by directly manipulating attention across features and by introducing exogenous variation in cued representations.
Partial Equilibrium Thinking, Extrapolation, and Bubbles
(with Paul Fontanier). Online Appendix. May 2025.
Review of Financial Studies, Accepted.
Abstract. We develop a dynamic theory of “Partial Equilibrium Thinking” (PET), which micro-founds time-varying return extrapolation: extrapolative beliefs are present at all times, but only sometimes manifest themselves in explosive ways. We formalize the distinction between normal times shocks and “displacement shocks” (Kindleberger, 1978), and study their interaction with extrapolative beliefs. In normal times, PET generates constant extrapolation and momentum. Following a displacement shock that increases uncertainty, PET leads to stronger and time-varying extrapolation, triggering bubbles and endogenous crashes. Our theory sheds light on both normal times market dynamics and Kindleberger’s narrative of bubbles within a unified framework.
Publications
Expectations and Learning from Prices
(with Paul Fontanier). Online Appendix. January 2024.
Review of Economic Studies, 2024.
Abstract. We study mislearning from equilibrium prices, and contrast this with mislearning from exogenous fundamentals. We micro-found mislearning from prices with a psychologically founded theory of “Partial Equilibrium Thinking” (PET), where traders learn fundamental information from prices, but fail to realize others do so too. PET leads to over-reaction, and upward sloping demand curves, thus contributing to more inelastic markets. The degree of individual-level over-reaction, and the extent of inelasticity varies with the composition of traders, and with the informativeness of new information. More generally, unlike mislearning from fundamentals, mislearning from prices i) generates a two-way feedback between prices and beliefs that can provide an arbitrarily large amount of amplification, and ii) can rationalize both over-reaction and more inelastic markets. The two classes of biases are not mutually exclusive. Instead, they interact in very natural ways, and mislearning from prices can vastly amplify mislearning from fundamentals.
Work in Progress
Expectations and Inelastic Markets