Economic transition and college premium in urban China

China has recently undergone a series of reforms on higher education aimed at raising the level of human capital. However, relatively little is known about returns to higher education (i.e. college premium) and how it varies across population in China over time. In this paper, we contribute to the literature by examining college premium in urban China during the period of 1995–2002. We also explore the differences in college premium by gender and by sector. To isolate the causal effects of college education, we employ a novel instrumental variable approach based on heteroskedasticity in college decision to circumvent potential endogeneity and measurement error problems. Our results imply a much larger college premium and a greater growth in the college premium relative to their OLS counterparts, consistent with the general literature on returns to education. More interestingly and starkly contrasting to the existing studies using OLS in the Chinese context, our results imply that (1) the college premium is larger for women than for men in the early stage of economic reforms, but the difference decreases and becomes statistically insignificant over time; (2) the college premium is larger for workers in the state-owned enterprises than that in the non-state-owned enterprises during the early stage of economic reforms; this gap is, however, reversed in the later stage. We discuss potential reasons for these results.