China’s massive expansion of higher education has transformed young people’s educational opportunities and labour market outcomes. Using data from the Chinese Household Income Project (CHIP) in 2013 and 2018, we estimate the causal effect of higher education expansion on graduates’ earnings with a Propensity Score Matching and Difference-in-Differences (PSM-DID) model. We consider students’ unobserved talents and construct a counterfactual framework, categorizing students by changes in their chances of attending college as always-takers, compliers and never-takers. The expansion policy significantly increases compliers’ earnings relative to never-takers, especially for those taking 4-year degrees. However, there is little evidence that always-takers gain a similar benefit. We also examine gender and rural – urban differences and find that the expansion policy favours disadvantaged groups, with women and rural compliers both enjoyed a larger college premium than their counterparts.