Job Openings and Labor Turnover Survey

The Job Openings and Labor Turnover Survey (JOLTS) is the Bureau of Labor Statistics monthly measure of US labour market flow, tracking job openings, hires, and separations across the nonfarm economy. Its February 2026 data, released March 2026, recorded a 3.1% hiring rate, the lowest since April 2020 at the onset of the pandemic. That reading became the starting point for Stanford Digital Economy Lab's April 2026 analysis, which applied the 0.6 percentage point decline from the 2023 baseline to the 158.6 million nonfarm workforce and calculated that AI is suppressing roughly 950,000 to 1 million annual hires.

JOLTS is distinct from the headline unemployment rate: it measures flows, not stocks. A declining hiring rate shows that employers are hiring less frequently without necessarily cutting existing staff: the dominant pattern Stanford identifies as the primary AI displacement channel. By contrast, Challenger, Gray & Christmas tracks declared redundancies, which represent the visible surface of displacement. The 34-to-1 ratio Stanford derives from these two sources is the gap between what is visible and what is happening.

On 14 May 2026, the New York Fed published an occupation-level study of job-posting data that directly contested Stanford's use of JOLTS. The NY Fed researchers found that the relative decline in postings for AI-exposed roles began before ChatGPT shipped in December 2022, and that junior and senior roles fell at similar rates, removing the timing evidence and the under-25s concentration that Stanford's causal story requires. That finding means the same JOLTS data now supports two opposed policy conclusions: Stanford reads it as evidence AI is suppressing a million hires a year; the NY Fed reads occupation-level postings and finds AI is not the cause at all. Until the Bureau of Labor Statistics publishes its delayed GenAI workplace paper, JOLTS remains the primary federal input to the displacement argument while the causal interpretation is contested.