The Pulse
Warwick Study Links Remote Work to Junior Hiring Declines
A University of Warwick working paper analyzes 243 million new hires and 407 million job postings across four countries. Its authors find that remote-work exposure better predicts weaker junior hiring than generative AI exposure when the tw
AI.info Team ·
The University of Warwick study analyzes 243 million new hires and 407 million online job postings across the United States, United Kingdom, Canada and Australia—and finds that remote-work exposure better predicts declines in junior hiring than generative AI exposure when the two are examined together. The data cover 2017 to 2025. The authors say the result challenges claims that AI alone explains why employers have been hiring fewer early-career workers.
Peter John Lambert and Yannick Schindler, the paper’s authors, compare how exposure to working from home and generative AI relates to the share of new hires who are junior workers, and to job ads seeking applicants with limited experience. Their paper is a working paper, not peer-reviewed research, according to the Warwick Research Archive.
Remote-work exposure holds up in the comparison
When the researchers assess each exposure separately, a two-standard-deviation increase in either one predicts a decline of about five percentage points in the junior share of new hires by 2025. Each also predicts roughly a three-point drop in the share of job postings requiring limited experience. Those estimates alone cannot distinguish whether the decline tracks AI, remote work or both: the occupations most exposed to one are often exposed to the other.
“Estimated jointly, the WFH effect remains, while the GenAI coefficient attenuates sharply and is often statistically indistinguishable from zero.”
Peter John Lambert and Yannick Schindler, “The Broken Ladder: AI, Remote Work, and Early-Career Hiring”
The researchers also test firms’ actual remote-work adoption. They compare organizations whose job ads offered remote or hybrid arrangements in 2021–22 with similar firms that did not, then examine junior hiring and experience requirements in 2023–25. Their analysis finds that the firms offering those arrangements went on to reduce the junior share of hiring relative to the comparison firms.
Four countries, two large data sets
The authors build one data set from employer–employee-linked records assembled from résumés and another from online vacancies. They define junior hiring by the positions filled and workers’ employment histories; for job ads, they track postings requiring no more than three years of relevant experience. The paper uses difference-in-differences designs at the occupation, region and firm levels.
The authors’ explanation is organizational as well as technological. Early-career employees often need supervision, feedback and time to learn on the job; remote arrangements, they argue, can make that support harder to provide. If training becomes more difficult, employers may have less reason to hire people who still need it.
The findings do not rule out AI’s effects
Lambert and Schindler do not claim generative AI has no effect on junior employment. They say their results concern the relative hiring of junior and senior workers through 2025, and that exposure-based comparisons can confuse AI exposure with remote-work exposure when they do not measure both directly. Their paper also reports tests using alternative exposure measures and flexible controls, but its conclusions remain estimates from observational data rather than proof that remote work caused every hiring decline.
The distinction matters for employers weighing how to respond. The authors argue that their results point toward improving supervision and training in remote or hybrid settings, rather than treating office attendance as the only remedy. The paper’s data end in 2025, leaving the relative contribution of actual AI adoption and remote-work practices in later hiring trends unmeasured.