Consider the following facts. In 1950, the richest countries attained an average of 8 years of schooling whereas the poorest countries 1.3 years, a large 6-fold difference. By 2005, the difference in schooling declined to 2-fold because schooling increased faster in poor than in rich countries. What explains educational attainment differences across countries and their evolution over time? We consider an otherwise standard model of schooling featuring non- homothetic preferences and a labor supply margin to assess the quantitative contribution of productivity and life expectancy in explaining educational attainment. A calibrated version of the model accounts for 90 percent of the difference in schooling levels in 1950 between rich and poor countries and 71 percent of the faster increase in schooling over time in poor relative to rich countries. These results suggest an alternative view of the determinants of low education in developing countries that is based on low productivity.