How important is the differential productivity of primary schooling across countries in explaining international differences in human capital? In this paper, I presents the first micro-econometric evidence on this issue using a unique child-level panel dataset with linked assessments of quantitative ability at 5 and 8 years of age from four developing countries – Ethiopia, India, Peru and Vietnam. I first document that although some cross-country gaps in test scores are evident already at 5 years of age, prior to school entry, these gaps grow substantially in the first years of schooling: children in Vietnam consistently score highest on average followed by Peru, India and Ethiopia respectively. Using value-added estimates, together with an RD/IV approach using discontinuities in grade completion arising due to month of birth and country-specific enrolment guidelines, I document sizable cross-country differences in the productivity of a school year. School year productivity, measured as learning-gains-per-grade-completed, varies from 0.45 SD-per-year in Vietnam to, for example, 0.2 SD-per-year in Peru. This differential productivity of a year in primary school, rather than differences in child endowments, nutrition, background characteristics or time use, largely explains the early divergence between Vietnam and other countries: equalizing this parameter to Vietnamese levels, leaving all endowments including learning at 5 years unchanged, closes the entire achievement gap between Vietnam and Peru, and 70% of the gap between Vietnam and India, at 8 years of age. (Author's abstract)