Information technology, government policy, and education mobility

Abstract

Mobility in education is essential to social mobility. Using data from an online education platform, we aim to detect how students’ relative ranking changes with adoption time, and examine the effects of student efforts, teacher efforts, and parent efforts respectively. The main result shows that adopting this platform prompts the convergence of relative ranking within a class, improves the ranking of students with low initial rankings, and lowers the ranking of students with high initial rankings. The results also show different effect on relative rankings from efforts of different roles. In general, student effort significantly improves their performance, with especially larger effect for students with low initial rankings. Parental effort also significantly improves their children’s performance, with especially larger effect for students with medium initial rankings. Our heterogeneity results show that these effects are larger for male students (compared with female students) and rural students (compared with urban students). We then use mandatory social isolation in China in the first quarter of 2020 to examine the impact of fully online education on student academic performance. Using new data from a large online education platform, we find that students who rank lower in the class are more likely to benefit from fully online courses, although fully online education may have some negative effects on students who have performed well in the past. These effects were more significant for elementary school students, male students, and students living in urban areas. We then examined student behavior and found that our main findings could be explained to some extent by the fact that students who ranked lower invested more time and interacted more positively within the class. We also use the Double Reduction Policy in China as a shock and try to measure students’ performance before and after the policy shock and figure out the possible effect of the policy boosting educational mobility. With newest data from the online platform, we find a significant convergence of rankings of students in a class after the policy shock. Considering the problem of possible natural trends, we then use a difference in difference method to measure the effect of policy on education mobility to remove trends that may affect our results. Since the Double Reduction Policy may also affect those cities outside the experimental zone of the policy (as they view it as a signal), we don’t give a precise estimation of the pure treatment effect of the policy, instead, we examine the casual effect of between a heavily-treated sample and a slightly-treated sample on education mobility. Main difference in difference results show that generally the convergence of rankings of students in a class is more significant in heavily-treated sample zone, implying more education mobility after Double Reduction Policy proposed. Since this influential policy just came out for several months, our study on this topic is more likely an exploratory work, and hopefully we can estimate the precise casual effect of the Double Reduction Policy on students based on more sample and behavior data later.