corporate artificial intelligence activity and accounting information comparability: evidence from CEO generational differences
Accounting information comparability reflects not only common accounting standards but also the information-production process through which firms translate economic events into accounting outcomes. Using 12,355 firm-year observations for nonfinancial Chinese A-share listed firms from 2012 to 2022, we examine the relation between corporate AI activity and accounting information comparability. Our primary explanatory variable is the one-year-lagged standardized measure of ln(1 + AI patent applications)identified from patent classification codes, which captures firms’ verifiable AI technological capability. Firms with stronger AI capability exhibit greater accounting information comparability. Mechanism tests show that earlier AI activity is followed by stronger internal control and lower earnings management, both of which are associated with higher comparability. CEO cohort further shapes the organizational conversion of technological capability: the AI effect is stronger in firms led by CEOs born in 1970 or later. To assess the construct validity of the patent-based measure, we triangulate evidence from patents related to accounting information production, AI-related software and hardware investment, and AI signals in MD&A and annual reports. The relation between AI capability and comparability is stronger when firms make more explicit AI-related resource commitments and exhibit greater organizational attention to AI. The stronger effects among firms with greater analyst attention and higher internal coordination demands further point to the roles of external demand for comparable information and internal information-processing needs. The main findings are robust to alternative measures, double machine learning, a policy-exposure instrumental-variable design, and entropy balancing. Overall, the evidence shows how AI technological capability can translate into more comparable accounting information through information production, organizational embedding, and managerial influence.
Authors
- Chunli Wang
- Qianchao Li (ORCID: https://orcid.org/0009-0005-3488-2616)
- Linrong Wu
- Juping Li (ORCID: https://orcid.org/0009-0006-0890-9245)
Institutions
- Guangxi Normal University (CN)
- Guilin University
Publication Details
- Journal
- Scientific Reports
- Published
- 2026-09-29
- DOI
- https://doi.org/10.1038/s41598-026-69633-w
- Primary Topic
- Auditing, Earnings Management, Governance
- Type
- article
- Field-Weighted Citation Impact
- 0.00