Does Artificial Intelligence Shape Within-Firm Wage Inequality?
Ao Li · Mendeley Data · 12 Oct 2026
Read original sourceOpen access
Where this sits in the debate
- Will AI widen inequality? · its finding goes against the claim (why studies differ)
Direction of the finding as read from the abstract by an AI model (AI-written).
Key takeaway AI-written
A replication package for a study of AI adoption and within-firm pay gaps in Chinese listed firms from 2007 to 2023, measuring AI use from annual-report keywords and using national AI pilot zones in a difference-in-differences design. The paper's hypothesis is that AI narrows the gap between management and ordinary-employee pay by shifting labour demand toward high-skilled workers.
Abstract
This replication package accompanies the paper “Does Artificial Intelligence Shape Within-Firm Wage Inequality?”, which studies how the adoption of artificial intelligence (AI) affects wage inequality inside Chinese firms. The central research hypothesis is that AI adoption changes a firm’s internal labor demand and pay structure, and that this change narrows the wage gap between management and ordinary employees. More specifically, the paper hypothesizes that AI adoption reduces within-firm wage inequality primarily by shifting labor demand toward high-skilled workers, thereby altering the relative compensation of different employee groups. The analysis is based on panel data on Chinese A-share listed companies from 2007 to 2023. The main dependent variable is a firm-level pay gap, defined as the absolute difference between average management compensation and average ordinary-employee compensation. A smaller value therefore indicates lower within-firm wage inequality. The key independent variable is lnai, the log of AI-related keyword frequency in firms’ annual reports. This is a text-based measure of AI application: firms that discuss AI more intensively in their annual reports are interpreted as having greater AI adoption or AI-related activity. The DID treatment variable is treat_post, the interaction between treatment status and the post-policy period. The policy context is the National AI Innovation Application Pilot Zones (NAAPZ), using official lists from the Ministry. The DID design compares firms in pilot zones before and after the policy with firms not affected by the policy, conditional on fixed effects and controls. The empirical analysis is organized around several tables and Stata do-files. Table 1 reports descriptive statistics for the main variables. Table 2 presents baseline OLS regressions. Column (1) includes industry and year fixed effects; column (2) includes firm and year fixed effects. Table 3 reports the main DID results. Table 4 presents th
About this publication
Academic journal. Published in an academic journal — typically peer-reviewed before publication, though indexing alone doesn't guarantee review depth.
- Published
- 2026 (recently)
- Citations
- 0 (OpenAlex)
0 citations recorded by OpenAlex — too few 2026 papers in this collection for a fair comparison.
Related by theme, not necessarily by method or conclusion.
- The Effects of Generative AI on High Skilled Work: Evidence from Three Field Experiments with Software Developers
SSRN · 2024 - GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models
arXiv · 2023 - Experimental evidence on the productivity effects of generative artificial intelligence
Science · 2023 - The simple macroeconomics of AI
Economic Policy · 2024
Overview assembled automatically from collected metadata. How this site is built.