Working paper
Dissecting Corporate Culture Using Generative AI – Insights from Analyst Reports
SSRN
09/01/2023
DOI: 10.2139/ssrn.4558295
Abstract
Our study is among the first in finance, accounting, and economics to apply generative AI models as reasoning agents on analyst reports to gain insights into analysts’ views of corporate culture. We employ generative AI (ChatGPT) to analyze 2.4 million analyst reports between 2000 and 2020. Generative AI organizes analysts’ views into a knowledge graph that links different cultural values to their perceived causes and effects. In terms of influencing factors for corporate culture, analysts identify business strategy as a key factor for a large number of cultural values – customer-oriented, operations-oriented, innovation, results-oriented, adaptability, integrity, and risk control; and management team for teamwork, innovation, results-oriented, adaptability, integrity, risk control, and people-oriented values. In terms of business outcomes shaped by corporate culture, analysts identify innovation and adaptability as affecting almost all aspects of business operations, ranging from market share and growth to corporate ESG practices, while other values such as customer-oriented, operation-oriented, or people-oriented have less impact on business outcomes. We further provide evidence that analysts’ views of culture are distinct from values presented on corporate websites, and/or from the views of executives and employees. Finally, we show that analysts’ views of corporate culture are reflected in their stock recommendations and target price forecasts as well as impact price reactions to the release of their reports. We conclude that analysts’ research on corporate culture offers new insights into its causes and effects, and that we are closer to establishing the culture-firm value link.
Details
- Title: Subtitle
- Dissecting Corporate Culture Using Generative AI – Insights from Analyst Reports
- Creators
- Kai LiFeng MaiRui ShenChelsea YangTengfei Zhang
- Resource Type
- Working paper
- Publisher
- SSRN
- DOI
- 10.2139/ssrn.4558295
- eISSN
- 1556-5068
- Number of pages
- 88 pages
- Language
- English
- Date posted
- 09/01/2023
- Date updated
- 02/16/2024
- Academic Unit
- Business Analytics
- Record Identifier
- 9984701812702771
Metrics
1 Record Views