Market Premium, Social Process, and Statistical Naivety: Further Evidence on Differential Returns to Education in Urban China

In an article published in the American Sociological Review (Wu and Xie, 2003), we addressed the substantive question of whether or not higher earnings returns to education in the market sector than in the state sector in reform-era China are caused by the market mechanism. Developing a typology of workers based on their labor market histories, we proposed a model of selective transition of workers from the state sector to the market sector as an alternative explanation for higher earnings inequality and higher earnings returns to education in the market sector. Our main empirical results showed that later market entrants have significantly higher returns to education than state sector stayers, whereas we found insufficient statistical power in the data with which to differentiate the education returns of early market entrants from those of either later entrants or stayers

In this paper, we address comments made by Jann (2005) about our article and his naive use of statistical tests in evaluating our findings. We reconceptualize the problem and illustrate the power of causal analysis using the propensity score method, which depends on the ignorability assumption. In considering two groups – early and late entrants to the market sector – our results suggest a strong selection mechanism at work. That is, when workers with a low latent propensity of making a transition indeed did make a late transition to the market sector, they benefited the most from the transition. In other words, the story is more about under what conditions workers migrated to the market sector than how workers benefit from migrating to the market sector.