As artificial intelligence accelerates its penetration of China’s white-collar economy, a pair of leading economists are sounding the alarm over the disproportionate impact on young professionals and proposing a structural solution: a dedicated AI Employment Compensation Fund. The proposal, detailed in a research paper by Li Jia, Dean of the School of Economics at Singapore Management University, and Zhang Dandan, Vice Dean of the National School of Development at Peking University, argues that existing labor market institutions are ill-equipped to handle the specific disruptions that AI is generating and that a new, targeted financial mechanism is urgently needed.
AI Is Closing Off Entry-Level Career Pathways
The research draws on a rich empirical foundation: an analysis of approximately 1.63 million job postings scraped from Zhaopin, one of China’s largest recruitment platforms. From this dataset, the economists constructed an “AI-LLM Occupational Exposure Index” that quantifies the degree to which different job categories are susceptible to automation by large language models.
The findings, presented in their Caixin article and reported on by Fred Gao on Substack, paint a concerning picture. White-collar jobs are heavily concentrated in occupations with high AI exposure. More troublingly, the research reveals a perverse dynamic: the higher the occupational exposure risk, the higher the entry barriers employers set for junior-level applicants. In other words, as AI takes over routine cognitive tasks, employers are demanding more from the human workers they do hire, raising the bar precisely for the young, inexperienced workers who are least able to clear it.
This dynamic is effectively closing off the traditional on-ramps through which generations of Chinese workers have entered the professional workforce. As EastFrontier has previously reported, the government has been aware of this trend for some time, but policy responses have so far been limited in scope. The retraining programs announced earlier this year have been met with skepticism from many young workers who question whether new skills will be sufficient in an environment where AI capabilities are advancing faster than human retraining can keep pace.
The Macroeconomic Externality of Automation
Li and Zhang’s analysis goes beyond the individual level to identify a systemic macroeconomic risk. They argue that AI-driven labor substitution generates what they term an “aggregate demand externality.” When individual firms automate jobs, they capture the cost savings privately. However, they do not internalize the broader economic consequence: displaced workers consume less, which reduces aggregate demand across the economy and ultimately feeds back as weaker growth and fewer business opportunities for the very firms that automated in the first place.
This framing positions AI-driven unemployment not merely as a social welfare problem but as a structural economic risk that requires a structural economic response. It also provides the intellectual justification for a levy on the entities that benefit most from AI automation, a politically sensitive but economically coherent argument.
The “One Fund, Two Pillars” Framework
To address these challenges, the economists propose a comprehensive framework structured around “one fund, two pillars, and three supporting measures.” The centerpiece is the AI Employment Compensation Fund, which would be financed through a diversified set of revenue streams: fiscal allocations, surpluses from the existing unemployment insurance system, employer contributions, and potentially novel surcharges tied to data and computing power revenues.
The fund would be deployed for two primary purposes: retraining workers displaced by AI automation, and providing direct support for young and low-skilled workers who are struggling to find entry-level positions in high-exposure occupations.
The “two pillars” of the framework address both immediate and structural needs. The short-term pillar includes the establishment of an AI-exposure-based employment risk monitoring and early warning system, the promotion of flexible working hours, and targeted income support for workers in high-risk industries. The long-term pillar envisions a more fundamental transformation of China’s human capital infrastructure: a lifelong learning system built around portable “skills accounts” and “learning accounts” that workers can carry across employers and industries. Crucially, the economists also call for formally incorporating AI employment governance into China’s Employment Promotion Law and future AI-related legislation, giving the framework legal teeth.
China’s Path in Global Context
The paper situates China’s emerging approach within a comparative global framework, contrasting four national models. Singapore represents a state-led coordinated transformation, with active government management of the labor market transition. The EU has opted for procedural constraints through the AI Act, focusing on transparency and accountability. The U.S. has largely retreated from federal coordination, leaving a fragmented landscape of state-level regulation. China, the paper argues, is developing a distinctive “governance-within-development” path, one that seeks to enable AI innovation while managing its social consequences through targeted interventions.
This framing is significant because it positions the proposed fund not as a brake on China’s AI ambitions, but as a complement to them. As Chinese courts have already ruled that AI adoption cannot justify firing workers, the legal and policy environment is clearly moving toward a more active role for the state in managing the labor market consequences of automation. The AI Employment Compensation Fund proposal represents the most comprehensive articulation yet of what that active role might look like.