| dc.description.abstract | This dissertation investigates how peers affect students’ educational and non-educational out-comes or how pension eligibility has an effect on retirement and healthcare utilization. For the causal identification, I rely on as-good-as random or quasi-experimental variation in the treatment.
First, I examine the persistent effects of ordinal rank by using idiosyncratic variation in the test score distribution across classes. To avoid the self–selection problem, I exploit longitudinal data from middle schools in South Korea, where students are randomly assigned to classrooms. I find a consistently positive impact of students’ ordinal rankings in 7th grade on their test scores even five years later. I also find that a higher math rank in 7th grade will increase the likelihood of choosing the math–science track, taking higher-level math courses, and being interested in obtaining a STEM degree. The results of postsecondary outcomes indicate that ordinal rank in 7th grade is positively associated with the probability of attending a university. The results also show that a higher math rank in 7th grade increases the likelihood of applying to a STEM major as well as majoring in a STEM field while attending a university. These effects are likely driven by changes in students’ self-confidence and effort provisions, parental investment, and students’ attendance in a preferred school.
In addition, I study the spillover effects of misbehaving boys on others’ behavioral outcomes, leveraging the random assignment of students into classrooms in South Korean middle schools. Peers with single parents are more likely to misbehave in class, providing an instrument for peers’ misbehavior to overcome the reflection problem. Misbehaving boys lead to an increase in the intensity of other students’ misbehaviors in the classroom, whereas misbehaving girls do not have statistically significant effect on others. The heterogeneous analysis suggests that these effects are more salient for male pupils and interactions between male students who live in low-income households.
Lastly, using the exogenous rule for the public pension benefit and a dataset for the elderly population from South Korea – one of the developed Asian countries, I provide new evidence of how pension eligibility affects retirement and healthcare utilization. To overcome selection bias, I rely on a regression discontinuity design (RDD) to compare the outcomes of those barely above and below eligibility age thresholds. The findings indicate that crossing pensionable ages increases the probability of being in retirement and inpatient care utilization, while doing so has no significant effect on outpatient care utilization. The local average treatment effect (LATE) of retirement on healthcare utilization is consistent with previous studies in developing country or some high-income countries finding that retirement has a positive impact on inpatient care utilization. However, an additional analysis relying on the Intent-to-Treat (ITT) parameters for covariates-based subgroups suggests that retirement might not be a strong or credible driver of the changes in inpatient care utilization across all subgroups. | |
