This paper assesses the factors underpinning trends in mathematics learning for Grade 4 pupils in Ethiopia based on data collected in 2012-13 (the Young Lives surveys, YL) and 2018-19 (the RISE surveys). It combines comparable data on attainments on tests of students’ mathematics knowledge with information on their family background, their teachers, and the schools they attend. The period covered by the study encompasses an education reform, the General Education Quality Improvement Program – Phase II (GEQIP-II). GEQIP-II’s goals included increasing access to primary education and the quality of education that was provided.

We find that mathematics teachers’ educational qualifications and teacher content knowledge in mathematics improved between 2012 and 2018. Despite this, students in 2018 have learning levels lower than students in 2012. This reflects, in part, changes in the composition of the samples, with the 2018 sample coming from poorer and rural households. However, students in 2018 show greater learning progress over the course of a school year compared to students in 2012. Using a value-added model, we show that learning outcomes in mathematics at the end of the school year are associated with teacher content knowledge, that this association has increased in magnitude over time, and that this content knowledge is especially important for weaker students. The key to reconciling these paradoxical findings is, thus, to recognize that the reforms that underpin these improvements in school quality may well be working in multiple ways, changing the composition of the student body who attend school as well as increasing the amount of learning that takes place in the classroom, particularly for weaker students. This, juxtaposed with the fact that the education reforms implemented over the last decade have sought to both include students from disadvantaged backgrounds and to improve teacher quality, is suggestive of the possibility that the GEQIP initiatives were producing positive learning results.