Geographic Heterogeneity of Returns to Education in China

Open Access
Jiang, Liyin
Area of Honors:
Bachelor of Science
Document Type:
Thesis Supervisors:
  • David Shapiro, Thesis Supervisor
  • Russell Paul Chuderewicz, Honors Advisor
  • returns to education
  • geographic heterogeneity
  • China
In this thesis, I look at the geographic heterogeneity of returns to education using data from the Chinese Household Income Project (CHIP) in 1995 and 2002. I begin with an earning equation with heterogeneous returns to education, based on Mincer's work (1974). However, we confront sample selection and endogeneity where we need Heckman's selection model (1979) and an instrumental variables (IV) method to solve. For the IV method, I use regional level ($\ln$) average income, ($\ln$) unemployment rate and ($\ln$) education expenditure as the instruments, test the endogeneity of education by a Durbin-Wu-Hausman (DWH) test, and then estimate returns to education. After comparing returns to education from using different models, I demonstrate the spatial autocorrelation (SAR) model and estimate the correlation coefficient of returns to education in different regions, using results from previous estimations. I find that the heterogeneity was stationary in 1995 but it displayed concavity in 2002 because dependence coefficients increase as educational level increase. This leads to the conclusion that higher level of education will reduce heterogeneity of returns to education, under the assumption that people are capable of searching for better paid jobs, which will be shown in the thesis.