AI Subsidies Cut Young Hiring 10%, Korean Study Finds

Subsidized AI adoption cut new hiring of workers aged 19 to 34 by 10.1% at small and midsize South Korean firms, according to a causal analysis of a government voucher program. Total new hiring at firms that received the subsidy fell 8% over three years.

Lee Hwan-woong, an economics professor at Konkuk University, presented the findings on Nov. 9 at a policy debate titled "Issues and Solutions in Employment and Labor Policy in an Era of Great Transition." The event was hosted by the Labor-Management-Public Interest Forum together with the Korean Economic Association of Labor, the Korean Industrial Relations Association, the Korean Society of Labor Law and the Korean Comparative Labor Law Association.

What did the study measure?

Lee tracked firms that participated in the AI Voucher Support Program run by the Ministry of Science and ICT. The program subsidizes purchases of AI solutions and consulting fees to strengthen the digital competitiveness of smaller companies.

Using corporate panel data from 2015 to 2024, his team ran a causal analysis of employment changes over the three years before and after firms received the subsidy. The design compared the 636 selected applicants out of 2,239 with the 1,603 rejected ones.

The numbers point in one direction. Firms that got the money and actually adopted AI cut new hiring of 19-to-34-year-olds — the age group concentrated in entry-level roles — by 10.1% over three years. Because young-worker hiring covers not just fresh graduates but also workers with limited experience changing jobs, the findings suggest the entry channel narrowed across the whole age group.

The mechanics are straightforward. When firms embed AI into production processes or internal systems, it quickly takes over entry-level tasks typically assigned to new or young employees: sorting documents, basic data processing and routine inspection.

Korea's labor market structure amplified the effect. Dismissals and workforce reductions are legally constrained, so companies cut new hiring rather than pushing out existing staff. They controlled total labor costs by leaving vacancies unfilled when workers retired or left, or by scaling back hiring plans from the outset.

What does this change for policy?

The core problem, in Lee's reading, is that the government's evaluation framework cannot see any of this. Performance measures for support programs typically count only total headcount at a given point in time. If existing staff stay on, the figures create the illusion that employment has been maintained — even as entry routes for young job seekers close.

Lee recommended changing employment impact assessment measures first, to prevent the contradiction of government support squeezing out young job seekers. New hiring should be tracked separately from job changes and departures, not just whether headcount holds steady. That means making "employment flow" measures (tracking hires against exits) mandatory, so the government can monitor directly whether entry opportunities shrink while overall employment holds up.

Training programs came in for criticism too. Programs that stop at basic education, Lee said, should give way to an "AI training-to-hiring track" modeled on the Ministry of Employment and Labor's main youth hiring support programs — one where young people gain experience in actual entry-level roles at companies and move on to employment.

He also flagged an asymmetry in tax and subsidy policy: hiring a person raises fixed costs such as social insurance contributions, while tax credits and subsidies concentrate on capital investment and automation.

"Given the delays workers face in finding new jobs and the burden of income gaps, the pace of automation may be excessively fast from a social standpoint," Lee said. "Even if we expect long-term gains in productivity and employment, we need mechanisms to cushion the shock of the transition."