Census papers: AI-exposed grads hit by 12% hiring drop

The two Census Bureau papers, both coauthored by Lee Tucker, trace a specific chain: ChatGPT's release in late 2022 was followed within months by a hiring collapse in the industries most exposed to AI, primarily software and information-technology work. The first paper, from April 2026, tracked matched employer-employee records and found that hiring of workers ages 22 to 24 fell sharply in those exposed industries while less-exposed industries held steady. Crucially, Tucker said the decline in hires — not people losing existing jobs — was the primary cause of the unemployment rate.

The second paper, from the following week, followed graduates of the most AI-exposed decile of college majors. Their odds of holding a job one quarter after graduation fell by five percentage points, and full-quarter initial earnings dropped 13% after ChatGPT's release. That earnings decline is roughly what economists would expect from graduating into a severe recession, but there was no recession — the rest of the labor market held up fine.

The papers show the adjustment took two forms: about half of the earnings loss came from graduates earning less within the industries that did hire them, and the other half from a shift into different industries, mainly lower-wage sectors like restaurants and retail. Meanwhile, the students themselves were slow to react: hiring data turned within months of ChatGPT's release, but enrollment data shows students didn't start abandoning computer science in visible numbers until three years later, with undergraduate computer science enrollment falling 8.4% in spring 2026. The broader labor-market picture reinforces the pattern — the New York Fed's tracker put the underemployment rate for recent college graduates at 42% in the second quarter of 2026 — but the two Census papers are the sharpest evidence that the entry-level door for AI-exposed majors has reopened only partway.