The paper examines the causal effect of education on time preference. To define our measure of time preference, we use responses to hypothetical questions involving intertemporal trade-offs from the Indonesian Family Life Survey. We instrument years of education with exposure to the Indonesian INPRES primary school construction program of the 1970s that caused exogenous variations in the cost of going to school. The local average treatment effect of the program was a 9 percentage point decrease in the probability of a female respondent choosing the most impatient response for every additional year of schooling. The results are robust to alternative definitions of the time preference measures but sensitive to changes in samples and specifications. The findings add to the evidence on the endogeneity of individual preference parameters that are often taken to be constant in neoclassical economics.