Order Matters: The Effect of Second-Wave Migration on Student Academic Performance in Northwest China

The migration of hundreds of millions of workers from the Chinese countryside to the city has created a generation of Left Behind Children (or LBCs), who now number more than 60 million. Existing studies have not consistently estimated the impact of parental migration on the academic performance of LBCs. Some studies suggest that remittance income could improve academic performance by easing liquidity constraints and increasing investment in children and their education; other studies claim that parental absence could harm academic performance by decreasing parental care and increasing the domestic responsibilities of the children left behind. Because of these trade-offs, the results of empirical studies that seek to measure the net impact of being left-behind on academic performance may be inconsistent because the relative strength of the income and parental care effects may be different for first-wave migration (that is, migration during the first period of time in which any—or one—parent migrates) and second-wave migration (i.e., migration when the remaining parent leaves the home and there is no parental care at home).

In this paper we examine how school performance changes before and after the second wave of parents out-migrate. We draw on a panel dataset of more than 5,000 students from 72 rural primary schools in rural Northwest China. Using a difference-in-difference (DD) approach, supported by a placebo test that tests the assumptions underlying the DD approach, we find that second-wave migration has statistically significant negative impacts on student performance. Importantly, second-wave parental migration is shown to have a more negative impact on academic performance than first-wave migration. Specifically, scores of standardized math test of students in second-wave migrant households decreased 0.08 SD (standard deviations). This fall in test scores for the children of second-wave migrant households is 0.07 SD more than students in first-wave migrant households. Such a result is consistent with the hypothesis that the negative effect of losing the last of a family’s parental care (as the second parent out-migrates) is greater than the positive effect of extra income that the parent generates (especially given the fact that the family was already receiving remittances from the first parent). Heterogeneous analysis indicates that the negative impact is most pronounced for those who are most susceptible to a decrease in parental care, namely a family’s oldest child (who would be expected to take on more parental duties when both parents had out-migrated) and students who live at home (as opposed to living in school as a boarder).