China experienced a near 5-fold increase in annual Higher Education (HE) enrolment in the decade starting in 1999. Using the China Household Finance Survey, we show that the expansion has exacerbated the large pre-existing urban-rural gap in educational attainment underpinned by the hukou (household registration) system. We then instrument years of schooling using the interaction of childhood urban hukou status and the timing of the expansion, which is analogous to a Difference-in-Differences estimator which uses rural students to control for any common time trend. The 2SLS estimates of 17% and 12% for men and women respectively are substantially larger than their OLS counterparts of 5% and 6%, both allowing for county fixed-effects. Our 2SLS results can be interpreted as a Local Average Treatment Effect (LATE), i.e. the average treatment effect of HE attendance on earnings for urban students who enrolled in HE as a result of the higher education expansion.