We analyze the relation between public, education-related infrastructure and the quality of education in schools using a case-study of the construction and implementation of two large public libraries in low-income areas in Bogotá, Colombia. We assess the impact of these libraries on quality of education by comparing results in national test scores (Saber 11º) for schools close and far from these libraries before (2000-2002) and after (2003-2008) the libraries’ opening. We find non-statistically different from zero differences that could be attributed to the libraries’ implementation. We also introduce Oaxaca-Blinder decomposition on Difference-in-Differences (DiD) estimates in order to assess if variation of traditional determinants of test scores for mathematics, verbal and science explain the result estimates. These results are robust to alternative specifications of DiD, a synthetic control approach and an alternative measure of distance.