This paper provides evidences of heterogeneous human-capital externality using CHIP 2002, 2007 and 2013 data from urban China. After instrumenting city-level education using the number of relocated university departments across cities in the 1950s, one year more city-level education increases individual hourly wage by 22.0 percent, more than twice the OLS estimate. Human-capital externality is found to be greater for all groups of urban residents in the instrumental variable estimation.
Glaeser, Edward L., and Ming Lu. "Human-Capital Externalities in China." National Bureau of Economic Research, August 2018.