Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

From Transcripts to Insights: Uncovering Corporate Risks Using Generative AI

Alex Kim, Maximilian Muhn, Valeri Nikolaev

TL;DR

This paper assesses whether generative AI, specifically GPT-3.5-Turbo, can extract firm-level political, climate, and AI-related risk exposures from earnings call transcripts and whether these AI-derived measures outperform traditional dictionary-based proxies. By constructing RiskSum_{it} and RiskAssess_{it} from transcripts and testing their predictive power for stock volatility, investment decisions, and mitigation actions, the study demonstrates that GPT-based risk assessments are highly informative and often superior to summaries or bigram measures. Moreover, AI-driven risk assessments prove especially valuable for emerging AI risk, show validity outside the model’s training window, and are priced in equity markets, implying practical usefulness for investors and policymakers. The findings highlight the ability of LLMs to synthesize contextual information with broad knowledge to generate actionable risk indicators, while also noting prompts and validation caveats inherent to AI-enabled analysis.

Abstract

We explore the value of generative AI tools, such as ChatGPT, in helping investors uncover dimensions of corporate risk. We develop and validate firm-level measures of risk exposure to political, climate, and AI-related risks. Using the GPT 3.5 model to generate risk summaries and assessments from the context provided by earnings call transcripts, we show that GPT-based measures possess significant information content and outperform the existing risk measures in predicting (abnormal) firm-level volatility and firms' choices such as investment and innovation. Importantly, information in risk assessments dominates that in risk summaries, establishing the value of general AI knowledge. We also find that generative AI is effective at detecting emerging risks, such as AI risk, which has soared in recent quarters. Our measures perform well both within and outside the GPT's training window and are priced in equity markets. Taken together, an AI-based approach to risk measurement provides useful insights to users of corporate disclosures at a low cost.

From Transcripts to Insights: Uncovering Corporate Risks Using Generative AI

TL;DR

This paper assesses whether generative AI, specifically GPT-3.5-Turbo, can extract firm-level political, climate, and AI-related risk exposures from earnings call transcripts and whether these AI-derived measures outperform traditional dictionary-based proxies. By constructing RiskSum_{it} and RiskAssess_{it} from transcripts and testing their predictive power for stock volatility, investment decisions, and mitigation actions, the study demonstrates that GPT-based risk assessments are highly informative and often superior to summaries or bigram measures. Moreover, AI-driven risk assessments prove especially valuable for emerging AI risk, show validity outside the model’s training window, and are priced in equity markets, implying practical usefulness for investors and policymakers. The findings highlight the ability of LLMs to synthesize contextual information with broad knowledge to generate actionable risk indicators, while also noting prompts and validation caveats inherent to AI-enabled analysis.

Abstract

We explore the value of generative AI tools, such as ChatGPT, in helping investors uncover dimensions of corporate risk. We develop and validate firm-level measures of risk exposure to political, climate, and AI-related risks. Using the GPT 3.5 model to generate risk summaries and assessments from the context provided by earnings call transcripts, we show that GPT-based measures possess significant information content and outperform the existing risk measures in predicting (abnormal) firm-level volatility and firms' choices such as investment and innovation. Importantly, information in risk assessments dominates that in risk summaries, establishing the value of general AI knowledge. We also find that generative AI is effective at detecting emerging risks, such as AI risk, which has soared in recent quarters. Our measures perform well both within and outside the GPT's training window and are priced in equity markets. Taken together, an AI-based approach to risk measurement provides useful insights to users of corporate disclosures at a low cost.
Paper Structure (28 sections, 6 equations, 13 figures, 16 tables)

This paper contains 28 sections, 6 equations, 13 figures, 16 tables.

Table of Contents

  1. Introduction
  2. LLMs' Theoretical Usefulness in Measuring Risk Exposure
  3. How does an LLM encode textual information?
  4. The Value of General Knowledge
  5. Methodology and Implementation
  6. Data
  7. Earnings Call Transcripts
  8. Capital Market Variables
  9. Economic Variables
  10. Descriptive Statistics
  11. Descriptive Statistics and Industry-Level Averages
  12. Time Series Variation
  13. Face Validity
  14. Variance Decomposition
  15. Capital Market Consequences
  16. ...and 13 more sections

Figures (13)

  • Figure 1: Figure 1. Time Trend in SK Telecom's Political Risk Exposure
  • Figure 2: Figure 2. Measuring Risks with Generative AI
  • Figure 3: Figure 3A. Industry Averages of PRiskAssess
  • Figure 4: Figure 3B. Industry Averages of CRiskAssess
  • Figure 5: Figure 3C. Industry Averages of AIRiskAssess
  • ...and 8 more figures