Summary: In this paper are interested not just in countries' average educational enrollment and attainment, for which there has been a great deal of examination both from official and academic sources, but in how educational attainment differs by household wealth within countries. How much schooling are children from poor households India, Brazil, or Kenya receiving, both absolutely and relative to the rich in the same country? Answering this question, especially in a way that produces valid comparisons across countries is hampered by the limited availability, difficulty of use, and comparability of household survey data. The Demographic and Health Surveys (DHS), having applied essentially the same survey instrument in 35 countries potentially overcomes these problems. One potential limitation of the DHS is that it lacks questions on household income or consumption expenditures, which are conventionally used as indicators of households' economic status. However, in a separate methodological paper the authors shows that an index constructed from the questions asked in the DHS about household assets and housing characteristics (e.g. construction materials, drinking water and toilet facilities) works as well, and arguably better, than consumption expenditures as a proxy for household long-run wealth. This finding allows us to use a comparable method, principal components, in constructing a ranking of households within each country. The 'poor' are simply defined as the bottom 40 percent in each country, so while levels of poverty are not comparable across countries, the rankings are constructed using a similar method. An analysis of this data on education and wealth reveals three key findings. First, very low primary attainment by the poor is driven by two distinct patterns of enrollment and dropout. There is a South Asian and Western/Central African pattern in which many of the poor never enroll in school.
Official, scanned versions of documents (may include signatures, etc.)