NY Fed Asks Whether Job Postings Reveal AI's First Labor-Market Effects

The Federal Reserve Bank of New York has turned its attention to a question that now sits at the center of every labor-market conversation: can we already see the impact of artificial intelligence in job postings, or is the technology's influence on hiring still too early to measure?

The analysis, published on the New York Fed's Liberty Street Economics blog under the title "Do Job Postings Show Early Labor-Market Effects of AI?", approaches the question through one of the most widely used high-frequency indicators available to economists: online vacancy postings. Researchers have leaned on posting data throughout the past several years precisely because it responds to demand shifts faster than payroll surveys or unemployment figures, which arrive on monthly or quarterly schedules and are subject to revision.

The New York Fed itself maintains a prominent postings-based indicator, and its economists have long argued that vacancies capture employer demand in near real time — a property that makes them a natural instrument for detecting structural change, including the kind that AI adoption might produce.

The stakes for workers are concrete. If AI is already reshaping demand, the earliest evidence would plausibly show up in three places: the text of postings, as employers add AI-related requirements or software mentions to role descriptions; the composition of postings, as firms post fewer vacancies for tasks that automation can handle; and wage benchmarks attached to newly posted roles, particularly in clerical, customer-service, and entry-level analytical functions that large language models can partially perform.

For individual job seekers, the analytical question translates directly into positioning. A candidate in an occupation where postings now routinely list AI tools alongside core skills faces a different market than one where the technology has yet to appear in vacancy text. The presence or absence of these signals in postings data offers a measurable way to distinguish between the two situations, rather than relying on anecdote or headline-driven anxiety about automation.

The Liberty Street Economics authors frame their inquiry as an empirical one — an examination of what the posting data does and does not yet show — consistent with the blog's practice of publishing data-driven analysis from the New York Fed's research staff. Readers should treat the piece as an early read rather than a verdict: postings measure employer demand, not employment outcomes, and a decline in postings for a given role does not by itself establish displacement.

What the analysis offers is a benchmark. As AI capabilities and corporate adoption both expand, subsequent readings of postings data can be measured against this baseline to determine whether the labor-market effects remain speculative or have begun to register in the numbers employers actually post.