We estimated the causal impact of vouchers for after-school education programs on children’s academic and behavioral outcomes using experimental data of middle and high school students. Our identification strategy relied on the random assignment of after-school education vouchers provided to children who suffered as a result of the Great East Japan Earthquake. We estimated the value-added models of various outcomes such as academic test scores (mathematics and Japanese language art) and non-cognitive skill measures (self-esteem and quality of life) of the children. We carefully treated potential biases due to sample attrition and the small sample property by employing the inverse probability weight for regression analyses. Our estimation results revealed that the assignment of vouchers had a positive and significant effect on the increase of mathematics test scores of high school students immediately, and of language art test scores of middle and high school students in one year. These results were robust to the fully non-parametric permutation tests. We found that the assignment of after-school education vouchers negatively affected the self-esteem scores of middle school students immediately, but this relationship became weak and insignificant in one year. We found that the assignment of vouchers was positively related to the quality of life measure, but these relationships were insignificant. We also estimated the effect of vouchers on the children’s study time, finding that the assignment of vouchers had a positive and statistically significant effect on study time, mainly at home. However, these positive effects were insignificant in the fully non-parametric permutation tests. This indicates that the assignment of vouchers improved the quality of the study environment of the children without increasing their study time, thus resulting in better academic outcomes.