Paper: arXiv 2511.15214

Authors: Giuseppe Matera

Abstract

Economic behavior is shaped not only by quantitative information but also by the narratives through which such information is communicated and interpreted (Shiller, 2017). I show that narratives extracted from earnings calls significantly improve the prediction of both realized earnings and analyst expectations. To uncover the underlying mechanisms, I introduce a novel text-morphing methodology in which large language models generate counterfactual transcripts that systematically vary topical emphasis (the prevailing narrative) while holding quantitative content fixed. This framework allows me to precisely measure how analysts under- and over-react to specific narrative dimensions. The results reveal systematic biases: analysts over-react to sentiment (optimism) and under-react to narratives of risk and uncertainty. Overall, the analysis offers a granular perspective on the mechanisms of expectation formation through the competing narratives embedded in corporate communication.

Complexity vs Empirical Score

  • Math Complexity: 3.5/10
  • Empirical Rigor: 8.5/10
  • Quadrant: Street Traders — practical and empirical, lighter on theory

Why this score: The paper introduces an innovative ’text-morphing’ counterfactual framework using LLMs to isolate textual narratives from quantitative data, offering highly novel insights into analyst behavioral biases. Empirical rigor is exceptionally strong, benchmarked against over 300 financial predictors and tested with robust statistical procedures, though mathematical proofs/derivations are minimal.

Research Flowchart

  flowchart TD
    A[Research Goal: Impact of Earnings Call Narratives on Analyst Expectations] --> B[Data Inputs: Transcripts & Quantitative Financial Data]
    B --> C[Methodology: Extracted Narratives via NLP/LLMs]
    B --> D[Methodology: Novel Text-Morphing LLMs]
    D --> E[Counterfactual Transcripts: Vary Topical Emphasis, Hold Quant Fixed]
    C --> F[Predictive Modeling & Bias Measurement]
    E --> F
    F --> G[Key Outcomes: Analyst Over-reaction to Sentiment Optimism & Under-reaction to Risk/Uncertainty Narratives]

Cited References (117)

  1. Comparing the Post–Earnings Announcement Drift for Surprises Calculated from Analyst and Time Series Forecasts (2006) — link
  2. Large Language Models as Simulated Economic Agents: What Can We Learn from Homo Silicus? (2023) — link
  3. Machine Learning as a Tool for Hypothesis Generation (2024) — link
  4. Narrative Asset Pricing: Interpretable Systematic Risk Factors from News Text (2023) — link
  5. Does Alternative Data Improve Financial Forecasting? The Horizon Effect (2024) — link
  6. Differences of Opinion and the Cross Section of Stock Returns (2002) — link
  7. Diagnostic Expectations and Stock Returns (2019) — link
  8. Man versus Machine Learning: The Term Structure of Earnings Expectations and Conditional Biases (2023) — link
and 109 more
  1. Noise in Expectations: Evidence from Analyst Forecasts (2024) — link
  2. Is There a Replication Crisis in Finance? (2023) — link
  3. Sticky Expectations and the Profitability Anomaly (2019) — link
  4. Which Subjective Expectations Explain Asset Prices? (2024) — link
  5. Simulating the Survey of Professional Forecasters (2024) — link
  6. Automated Social Science: Language Models as Scientist and Subjects (2024) — link
  7. Moving Targets (2024) — link
  8. Firm-Level Climate Change Exposure (2023) — link
  9. MPNet: Masked and Permuted Pre-training for Language Understanding (2020) — link
  10. Lookahead Bias in Pretrained Language Models (2024) — link
  11. Narrative Economics (2017) — link
  12. Entity Neutering (2025) — link
  13. Large Language Models: An Applied Econometric Framework (2025) — link
  14. From Generation to Judgment: Opportunities and Challenges of LLM-as-a-judge (2024) — link
  15. A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions (2024) — link
  16. Analysts’ Forecasts and Asset Pricing: A Survey (2016) — link
  17. Momentum strategies (1996) — link
  18. Expectations and the cross-section of stock returns (1996) — link
  19. The Market’s Mirror: Revealing Investor Disagreement with LLMs (2025) — link
  20. Firm Inattention and the Efficacy of Monetary Policy: A Text-Based Approach (2025) — link
  21. Attention to the Macroeconomy (2023) — link
  22. Earnings conference calls and stock returns: The incremental informativeness of textual tone (2012) — link
  23. Annual report readability, current earnings, and earnings persistence (2008) — link
  24. Management earnings forecasts and other forward-looking statements (2018) — link
  25. Corporate managers’ perspectives on forward-looking guidance: Survey evidence (2024) — link
  26. Firm-Level Political Risk: Measurement and Effects (2019) — link
  27. Does earnings guidance affect market returns? The nature and information content of aggregate earnings guidance (2007) — link
  28. Tone Management (2014) — link
  29. The Effects of Reporting Complexity on Small and Large Investor Trading (2010) — link
  30. Textual risk disclosures and investors’ risk perceptions (2013) — link
  31. The Sound of Uncertainty: Examining Managerial Acoustic Uncertainty in Conference Calls (2024) — link
  32. UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction (2018) — link
  33. Changes in Risk Factor Disclosures and the Variance Risk Premium (2023) — link
  34. Good-Bye I/B/E/S (or Not?) (2023) — link
  35. FinBERT: Financial Sentiment Analysis with Pre-trained Language Models (2019) — link
  36. The power of voice: Managerial affective states and future firm performance (2012) — link
  37. Why Language Models Hallucinate (2025) — link
  38. Approximately normal tests for equal predictive accuracy in nested models (2007) — link
  39. Predicting Future Earnings Changes Using Machine Learning and Detailed Financial Data (2022) — link
  40. Inefficiency in Analysts’ Earnings Forecasts: Systematic Misreaction or Systematic Optimism? (1999) — link
  41. A growing disparity in earnings disclosure mechanisms: The rise of concurrently released earnings announcements and 10-Ks (2019) — link
  42. Corporate Discount Rates (2025) — link
  43. Firms’ Perceived Cost of Capital (2024) — link
  44. Equity Return Expectations and Portfolios: Evidence from Large Asset Managers (2024) — link
  45. The return of return dominance: Decomposing the cross-section of prices (2025) — link
  46. Subjective Cash Flow and Discount Rate Expectations (2021) — link
  47. A model of investor sentiment (1998) — link
  48. Investor Beliefs and Expectation Formation (2025) — link
  49. The Natural Language of Finance (2025) — link
  50. Belief Overreaction and Stock Market Puzzles (2024) — link
  51. Bubble Beliefs (2025) — link
  52. Mental Models and Financial Forecasts (2024) — link
  53. Subjective Risk and Return (2024) — link
  54. Analysts’ Belief Formation in Their Own Words (2024) — link
  55. (Deep) Learning Analyst Memory (2024) — link
  56. Context-Based Interpretation of Financial Information (2024) — link
  57. Narrative Persuasion (2023) — link
  58. Narratives About the Macroeconomy (2023) — link
  59. Media Narratives and Price Informativeness (2023) — link
  60. The Macroeconomics of Narratives (2024) — link
  61. Assessing Look-Ahead Bias in Stock Return Predictions Generated By GPT Sentiment Analysis — → 2309.17322
  62. Measuring What is Top of Mind (2024) — link
  63. Subjective Models of the Macroeconomy: Evidence From Experts and Representative Samples (2022) — link
  64. From the Horse’s Mouth: Economic Conditions and Investor Expectations of Risk and Return (2013) — link
  65. What Have They Been Thinking? Home Buyer Behavior in Hot and Cold Markets (2012) — link
  66. Depression Babies: Do Macroeconomic Experiences Affect Risk Taking?* (2011) — link
  67. Predictability in financial markets: What do survey expectations tell us? (2009) — link
  68. Perspectives on Behavioral Finance: Does “Irrationality” Disappear with Wealth? Evidence from Expectations and Actions (2003) — link
  69. Expectations data in asset pricing (2023) — link
  70. Does the Stock Market Overreact? (1985) — link
  71. Psychology-Based Models of Asset Prices and Trading Volume (2018) — link
  72. Partial Equilibrium Thinking, Extrapolation, and Bubbles (2023) — link
  73. Expectations and Learning from Prices (2025) — link
  74. Resolving Estimation Ambiguity (2024) — link
  75. What Drives Very Long-Run Cash Flow Growth Expectations? (2023) — link
  76. Expectations Formation with Fat-Tailed Processes: Evidence and Theory (2022) — link
  77. Expected EPS × Trailing P/E (2024) — link
  78. Finance without exotic risk (2025) — link
  79. Asset Pricing with Fading Memory (2022) — link
  80. Five Facts about Beliefs and Portfolios (2021) — link
  81. Overreaction in Expectations: Evidence and Theory (2023) — link
  82. Expectations of Returns and Expected Returns (2014) — link
  83. Shaping liquidity: On the causal effects of voluntary disclosure (2014) — link
  84. Testing Asymmetric-Information Asset Pricing Models (2012) — link
  85. A Quantitative Analysis of Distortions in Managerial Forecasts (2024) — link
  86. Marketwide Memory (2025) — link
  87. Causal Narratives (2025) — link
  88. Factor Models, Machine Learning, and Asset Pricing (2022) — link
  89. The Ghost in the Machine: Generating Beliefs with Large Language Models (2025)
  90. What Hundreds of Economic News Events Say About Belief Overreaction in the Stock Market (2024) — link
  91. Asset Embeddings (2025) — link
  92. Memory and Beliefs in Financial Markets: A Machine Learning Approach (2024) — link
  93. Why Do Investors Like Short-leg Securities? Evidence from a Textual Analysis of Buy Recommendations (2025) — link
  94. Can ChatGPT Forecast Stock Price Movements? Return Predictability and Large Language Models (2023) — link
  95. Economic Representations (2025)
  96. Lazy Prices (2020) — link
  97. Redefining Financial Constraints: A Text-Based Analysis (2015) — link
  98. Product Market Synergies and Competition in Mergers and Acquisitions: A Text-Based Analysis (2010) — link
  99. Text-Based Network Industries and Endogenous Product Differentiation (2016) — link
  100. Credit Information in Earnings Calls (2023) — link
  101. The News in Earnings Announcement Disclosures: Capturing Word Context Using LLM Methods (2025) — link
  102. PEAD.txt: Post-Earnings-Announcement Drift Using Text (2023) — link
  103. Straight Talkers and Vague Talkers: The Effects of Managerial Style in Earnings Conference Calls (2017) — link
  104. When Is a Liability Not a Liability? Textual Analysis, Dictionaries, and 10-Ks (2011) — link
  105. Measuring Corporate Culture Using Machine Learning (2021) — link
  106. From Numbers to Words: Breaking Down Institutional Beliefs (2025) — link
  107. Time-Series and Cross-Section of Risk Premia Expectations: Evidence from Financial Analysts (2022) — link
  108. FOMC In Silico: A Multi-Agent System for Monetary Policy Decision Modeling (2025) — link
  109. The Value of Information from Sell-side Analysts — → 2411.13813

2 of 117 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.