How Do Heterogeneous Social Interactions Affect the Peer Effect in Rural-Urban Migration?: Empirical Evidence from China

In this paper, we use the "2002 Chinese Household Income Project Survey" (CHIP2002) data to examine how heterogeneous social interactions affect the peer effect in the rural-urban migration decision in China. We find that the peer effect, measured by the village migration ratio, significantly increases the individual probability of outward migration. We also find that the magnitude of the peer effect is nonlinear, depending on the strength and type of social interactions with other villagers. Interactions in information sharing can increase the magnitude of the peer effect, while interactions in mutual help in labor activities, such as help in housing construction, nursing and farm work in busy seasons, will impede the positive role of the peer effect. Being aware of the simultaneity bias caused by the two-way causality between social interaction strengths and migration, we utilize "historical family political identity in land reform" as an instrumental variable for social interactions. However, the hypothesis that probit and instrumental-variable probit results are not significantly different is not rejected. The existence of a nonlinear peer effect has rich policy implications. For policy makers to encourage rural-urban migration, it is feasible to increase education investment in rural areas or increase information sharing among rural residents. However, only an increase in the constant term in the regression, i.e., a "big push" in improving institutions for migration, can help rural Chinese residents escape the low equilibrium in migration.