In the urban labor market, the rural labor force can choose whether to become self-employed or work for wages depending on their stock of human capital and local labor market conditions. A simple Mincer earnings regression shows that the rate of return to schooling for wage earners is two percentage points higher than that for the self-employed. After correcting for bias in sample selection, the expanded Mincer earnings equation estimated the rate of return to schooling for wage earners at between 5.3 and 6.8 percent. From the standpoint of training, we found that the simplest form of training, short-term training and formal training played an important role in promoting migrant workers' repeat mobility. However, the simplest form of training did not have a significant effect on earnings, whereas short-term and formal training played an important determining role in this respect. Moreover, rights protection issues such as wage arrears also had an important effect on migrant workers' repeat mobility. In handling heterogeneity and endogenous educational variables among migrant workers, the authors found that the years of schooling of the parents of migrant workers were not an ideal instrumental variable.
