Abstract: Using valuation reports disclosing perceived cashflow growth (g) and discount rates (k) in M&A transactions, we examine what managers learn from target stock prices. Before correcting for endogeneity, both appear sensitive to prices—positively for g, negatively for k, and with equal magnitude—suggesting managers learn about both. However, using noise in prices as an instrument, only k reacts—with corrected estimates indicating that 89% of managers’ information about k comes from prices. Therefore, stock markets provide insights into risk and the compensation it requires, but not cashflows, which managers already understand well. Cross-sectional tests reinforce this conclusion.
Discussant: Ricardo Delao, University of Southern California
Abstract: We examine how large asset managers form and justify long-horizon beliefs using their Capital Market Assumptions (CMAs). Our evidence points to a common architecture: managers decompose equity return expectations into shared building blocks, populate them using heterogeneous modeling assumptions, and process information through causal narratives, while peer consensus anchors forecast revisions. Valuation change and growth explain 77% of cross-sectional dispersion and are most strongly linked to equity allocations. Valuation-change expectations are countercyclical, whereas growth expectations are procyclical, generating countercyclical return expectations overall, with substantial heterogeneity across managers. Disclosed modeling assumptions matter too: mean-reversion and historical calibration predict systematic deviations from peer consensus. Using a new LLM methodology, we extract directed, signed causal networks from CMA narratives. Greater network complexity and attention to valuation change are associated with underreaction to positive earnings news, whereas attention to dividend yield and downturns is associated with overreaction. Comparisons with N-CSR shareholder letters show that CMA narratives reflect persistent institution-specific investment views. Volatility and correlation forecasts, by contrast, vary less across managers and remain closely tied to historical realizations.
Discussant: Philippe van der Beck, Harvard University
Christian Heyerdahl-Larsen, BI Norwegian Business School
Philipp Illeditsch, Texas A&M University
Petra Sinagl, University of Iowa
Abstract: Recessions cause delayed declines in output, followed by recoveries with abnormally high growth. We develop an equilibrium model in which investors become aware of a new aggregate risk but must learn its severity from realized macroeconomic outcomes. The crisis-shape
process generates a gradual deterioration and recovery in output, while belief revisions about crisis severity raise risk aversion when outcomes are worse than expected. The model predicts that, after recession onset, risk premia and return volatility do not rise immediately but
instead increase gradually before peaking, generating hump-shaped dynamics unlike those in many non-expected-utility models (Ai and Bansal, 2018). When investors learn that the crisis is severe, this generates declining valuation ratios. These dynamics match empirical
patterns in output, risk compensation, volatility, and valuations around U.S. recessions.
Discussant: Aditya Chaudhry, Ohio State University
Abstract: We hypothesize that most individuals have little understanding of how the distribution of equityreturns evolves over horizons. As a result, individuals compress their estimates of near- (1-year)and long-term (10-year) uncertainty. Consistent with our hypothesis, individuals report estimatesthat imply an implausibly strong negative relation between risk and horizon when askedabout total returns but estimates that imply an implausibly strong positive relation betweenrisk and horizon when asked about average returns. The analysis helps explain puzzling resultsacross several literatures and has implications for household finance, corporate finance, and assetpricing.
Discussant: Xindi He, Georgia Institute of Technology
