Institute for Economic and Social Research

Vol. 90 | Seminar

2018-06-01

Topic: Peer Effects under Endogenous Selection: An Application to Local and Migrant Children in Shanghai Elementary Schools

Speaker: Associate Professor Chao Yang, Shanghai University of Finance and Economics 

Time: March 26th, 2018, 13:30–15:00

Venue: Conference Room 106B, Zhonghui Building (College of Economics, JNU)

Abstract:

We build a model to investigate peer effects in two kinds of schools, the public ones for both local and migrant children and the migrant ones only for migrants. In addition to allow variations in the intensities of influences from peers of different registered residential statuses (the so-called “hukou”), we also take into account endogenous school selection, as some unobservables may not only affect the performance of a child but also play a role in determining whether a migrant child can enroll in a public school or a migrant one. We adapt a switch regression model to the framework of social interactions under incomplete information, where the behavior of an individual can be affected by her expectation on peer’s outcomes conditional on publicly known covariates and school selection. Then, endogenous selection of one child can affect others’ outcomes through peer relations. With jointly normal disturbances, the mean of outcomes in classes of each school type will be affected by the inverse Mills ratio in a nonlinear way. Consistent estimates can be derived by either a 2- stage NLS or a 2-stage GMM methods, with the conditional mean solved as a fixed point in the inner loop nested in the second stage. Both approaches have nice small sample performances as Monte Carlo simulations show. In the empirical application to a survey data on elementary school children in Shanghai, we find that in public schools, a child is affected more intensively by classmates with the same registered residence status than the peer students of another status. In addition, a migrant has significant positive influence on a local child but not vice versa. When selection bias is assumed away, the influence from the migrant peer to a local child will be underestimated.


back

Copyright © 2019 Institute for Economic and Social Research ICP record No.: Yue ICP Bei No. 12087612