We evaluate a two-year accelerated education program in Niger by randomizing schools to villages and places in the schools within villages. The program had large effects on education and learning: 183 percent more treated children start lower secondary education (our main educational outcome), they are 121 percent more likely to be literate, and they score 75 percent higher on a numeracy index. We find no indications of spillover effects of the program. In particular, control children in treated villages are very similar to children in control villages at endline. Despite the large effects on educational and learning outcomes we find no effects on any of our other five main pre-registered outcomes: Attitudes toward the appropriate marriage age and gender equality, well-being, religious extremism, or engaging in hazardous work.

Written with Anne Kielland (FAFO), Andreas Kotsadam (The Frisch Centre, University of Oslo), and Jing Liu (FAFO)