To shed light on the effectiveness of educational inputs for student outcomes, this paper examines the effect of private tutoring expenditures on the academic performance of middle school students in South Korea. To address endogeneity, the paper uses instrumental variables, first-difference, propensity-score matching and nonparametric bounding methods. We apply these methods to a panel dataset from South Korea, the Korea Education Longitudinal Study. The results show that the true effect of private tutoring remains, at most, modest. Instrumental variables (first-difference) estimates suggest that a 10-percent increase in expenditure raises a test score by 0.03 standard deviations or 1.1 percent (0.002 standard deviations or 0.08 percent). Matching estimates imply that the same amount of increase in expenditure leads to a 0.33 to 0.72 percent higher average test score. The tightest bounds of the effect of tutoring reveal that a 10-percent increase in expenditure improves the test score by a low of 0 to a high of 2.01 percent, while statistical tests fail to rule out zero effects. The modest effects of private tutoring found in the present study are comparable to the effects of public school expenditures on test scores and earnings estimated in previous studies. [ABSTRACT FROM AUTHOR] .