In Working Paper CES-26-56, titled "Graduating into Disruption", the Center for Economic Studies at the U.S. Census Bureau presents an empirical assessment of the entry-level labor market. Published in mid-September 2026, the paper investigates how generative AI systems are reshaping employment and earning opportunities for recent college graduates in the United States. Rather than relying on speculative forecasts or corporate surveys, the study uses administrative records to trace concrete career starts. The findings reveal substantial disruptions in fields that were previously considered secure routes to high-earning professional careers.
The empirical foundation of the study is notable for its massive scale and administrative precision. Economists examined longitudinal records from the Post-Secondary Employment Outcomes and Longitudinal Employer-Household Dynamics databases, covering roughly 6.67 million bachelor's degree recipients. This administrative sample represents approximately 29 percent of all bachelor's degrees conferred across the graduation cohorts from 2016 through 2024. By tracking these cohorts over time, the researchers were able to rigorously compare outcomes between fields with differing levels of generative AI exposure.
The impact on initial hiring rates among technology-focused majors is immediate and severe. For the top 10 percent of college majors most exposed to generative AI, including computer science, software engineering, and information systems, hiring opportunities contracted sharply. Graduates from these programs experienced a 5 percentage point decline in their probability of securing employment in the first quarter after graduation compared to peers in low-exposure disciplines. The conventional transition from university coursework straight into formal corporate engineering roles has broken down for a measurable share of applicants.
The financial penalty for graduates who do find employment is even more pronounced. The paper finds that real starting salaries in the first post-graduation quarter plummeted by 13 percent for graduates in these highly exposed fields. The authors contextualize this drop by comparing its severity to graduating during a major macroeconomic recession. Experiencing such a pronounced decline in purchasing power immediately upon graduation sets a difficult financial baseline for young professionals as they enter the workforce.
A granular analysis of the wage data reveals that the 13 percent drop is driven by two distinct mechanisms of equal magnitude. Roughly half of the overall decline stems from lower initial wages paid within primary tech and engineering industries. The other half is attributable to a displacement effect, as graduates unable to secure technology jobs migrate into lower-paying service sectors, including retail and food service. This forced occupational downgrading substantially depresses the aggregate earnings of the entire graduate cohort.
These administrative findings underscore a stark divergence between enterprise automation and early-career economic reality. While technology firms rapidly automate internal development workflows, generative AI is creating measurable wage and employment dents for junior talent. Instead of spurring immediate entry-level hiring, generative tools appear to be replacing or devaluing tasks traditionally assigned to novice workers. The Census Bureau working paper provides concrete evidence that generative AI has begun to restructure the economics of entering the high-tech workforce.