This paper attempts to identify neighborhood peer effects on children's dropout decision in rural China using the China Health and Nutrition Survey data (CHNS). Identifying peer effect is complicated by several endogenity problems including “self-selection” problem, “reflective” problem and uncontrolled “correlated effect”. By taking advantage of the special feature of “Hukou” system and “son preference” phenomenon in rural China, the endogenity issues are quite reasonably addressed. More specifically, we discover a new and valid instrumental variable for peer's dropout rate: peers' firstborn boy rate. Intuitively, the more the firstborn boy peers, the lower the peers' dropout rate because of son preference and the lower one's own dropout probability due to peer effect. It is found that as peers' dropout rates increase by one percentage point, the child dropout rate would increase by 0.393 to 0.504 percentage points, the corresponding social multiplier effects of peer dropout are from 1.647 to 2.016. It is also found that elder kids and females are more susceptible to peer pressure in dropout decisions. Many other interesting findings are documented.
