SMU SOE Seminar Series (September 11, 2025): Measuring Occupational Exposure to AI Technologies in China and Its Impact on Labor Demand

TOPIC:

ABSTRACT

This paper measures occupational exposure to large language model (LLM) technologies and examines its impact on labor demand in China. We construct an index of LLM exposure using detailed task descriptions from 1.52 million online job postings between January 2018 and July 2025. We then estimate the effects of exposure on employment, educational requirements, and wages.

We find that exposure among newly created jobs has declined over time. Occupations with higher exposure are primarily white-collar jobs with higher educational requirements and wages, such as accountants, editors, sales staff, and programmers. Instrumental variable estimates at the occupation level show that higher LLM exposure is associated with lower labor demand, slower wage growth, greater within-occupation wage dispersion, and higher requirements for education and work experience.

These findings suggest that the substitution effect of AI technologies has already emerged in the Chinese labor market. The results point to the need for mechanisms to monitor labor market adjustments and to strengthen social protection, while encouraging firms and workers to create new positions to facilitate job transitions.

Click here to view the CV.