Work

Too Smart to List? AI Disruption, Financial Reporting, and Corporate Governance in the IPOs of Chinese Fintech Ventures: Evidence from an Agent-Based Computational Model

Year: 2026
Type: preprint
Abstract: Chinese fintech ventures adopt artificial intelligence (AI) faster than almost any other entrepreneurial population, yet the ventures that adopt the most AI do not list most often. We develop an agent-based computational model in which 5,000 fintech ventures, underwriters, a securities regulator, and IPO investors interact from 2013 to 2025 under the institutional conditions of China's approval and registration regimes. The model treats AI as an entrepreneurial actor that raises productive capability but also creates algorithmic opacity and reshapes the discretion embedded in financial reporting. Estimating archival-style models on the simulated data, we find an inverted U-shaped relationship between AI adoption intensity and the annual hazard of listing. AI disruption exposure raises pre-IPO discretionary accruals, and strong corporate governance cuts this effect roughly in half. Accrual magnitude raises first-day returns, income-increasing accruals predict lower three-year buy-and-hold returns and higher distress, and reporting quality fully mediates the effect of disruption on regulatory approval. Monte Carlo replication, global sensitivity analysis, specification curves, and measurement-error experiments show which of these inferences an archival researcher could reliably recover. Counterfactual experiments indicate that governance upgrades and explainable AI raise listing rates and post-IPO performance more than AI-enabled regulatory review does.
Author: jing zhang
Language: English
Subfield: Accounting
Open Access status: green
OpenAlex ID: W7235219339
Cites:
Cited by:
Related to: 10