Employing exogenous variations from random classroom assignment and field experiment, this thesis investigates three topics. The first two topics focus on peer effects on educational outcomes and bullying behaviours in China, whereas the third paper evaluates a subsidized employment program in North Macedonia. Chapter 2 measures the spillover effects of parental migration using the sample schools in which 7th-grade students are randomly assigned to classes during the compulsory schooling period. I find that having additional 10 percentage points of left behind children in the class led to a decrease in total test score of 5 percentiles. The medium run peer effect (one year after they met) is mainly driven by short-run peer effect taken place in about 10 weeks after they met. In Chapter 3, using the same strategy, I show that the within-class inequality of family socioeconomic status (SES) contribute to number of bullies in the class, as students from lower SES distribution want to gain peer status while these at the top want to secure their status. Policy simulations imply an intervention on class assignment may help the school to reduce the bullying incidents via reducing the bullies. Chapter 4 evaluates the impacts of the SEP experiment in which applicants were randomly selected to attend job interviews. It provides a wage subsidy to eligible employers for hiring an applicant. We find that attending the job interview led to more than 20 percentage point increase in employment and the effect persists at slightly lower percentage points afterwards. The programme’s effects are mainly concentrated among the most vulnerable groups.