Consequences of Private Tutoring for Educational Attainment: The Case of South Korea
This study attempts to reconcile different findings from previous studies by estimating the effect of private tutoring on various educational outcomes based on the same dataset. Data come from Korean Educational Longitudinal Study 2005-2014, which collected information on educational progress of children from middle school years to their early twenties. We estimate ordered logit models for GPA-based ranks, OLS regression models for college SAT scores, multinomial logit models for college enrollment results. However, the standard regression approach does not consider that parents and children who take private tutoring may be different from others in unobserved as well as observed characteristics. We use fixed effect approaches and instrumental variables techniques to address the endogeneity bias. Results from different analytic methods are compared. Implications for educational inequality and social mobility are discussed.