Paul Glewwe

One of Glewwe's main research areas concerns the relationship between health and education. For example, he (with Hanan Jacoby) finds early childhood malnutrition - and not borrowing constraints or the rationing of school places - to be the likely cause of delayed enrollment in primary school in Ghana,[8] and documents (with Jacoby and Elizabeth M. King) how malnutrition among young children in the Philippines impairs their academic achievements by delaying the age at which they enroll into school and causing them to learn more slowly, though not by decreasing their effort exerted at school (in terms of attendance, homework, etc.).[9] Looking into the link between child health and maternal education, Glewwe argues based on evidence from Morocco that mothers' health knowledge, which is generally correlated with (though not necessarily caused by) their schooling, is probably the main pathway how maternal education achieves its strongly positive impact on child health and nutrition in developing countries, which consequently suggests large public health payoffs to female health education in school.[10]

Since the early 2000s, Glewwe has used randomized controlled trials (RCTs) in order to investigate the impact of the quality of the supply of education, e.g. in terms of school supplies or the quality of teaching, on learning outcomes. For example, Glewwe finds (with Michael Kremer, Sylvie Moulin, and Eric Zitzewitz) that in Kenya neither the provision of textbooks nor of flipcharts was effective in raising average student scores and that generally only the best students were able to take advantages of the improvement in school supplies. This in turn raises the question of whether the emphasis of many developing countries' (strongly centralized) education systems on top-down improvements to the supply of education may be guided rather by elite bias than by a concern for broad increases in students' learning outcomes.[11][12] In another RCT, Glewwe (with Michael Kremer and Nauman Ilias) finds that rewarding primary school teachers in Kenya based on students' test scores and penalizing them based on their students not attending the exam leads teachers to increase the number of test preparation sessions (but not teacher attendance or homework assignments), which then increases students' test scores and exam participation, but is ineffective in reducing dropout rates.[13] Earlier on, having found school characteristics in Ghana to be highly correlated with student achievement, e.g. via average grade attainment, Glewwe (with Jacoby) had argued that improvements to school quality, such as repairs of classrooms, may be a cost-effective investment into education in Ghana relative to the provision of more teaching materials and better trained teachers.[14]

Glewwe's other findings include the following:

• That many pre-1990s estimates of rates of returns to schooling are significantly biased because they fail to account for differences in ability and school quality and, even if accurate, may provide poor guidance to education policy compared to rates of return to investments in school quality;[15]
• That the vulnerability of households in Peru to welfare decreases during macroeconomic shocks decreases in education and access to foreign transfer networks, increases if households are headed by women or have more children, and is hardly mitigated by Peru's social security programme (with Gillette Hall);[16]
• That the willingness to pay for schooling in households in rural Peru is high enough at all income levels to cover the operating costs of new schools in their villages, thus suggesting that increasing school fees to raise revenue for educational improvements in developing countries may be feasible (with Paul Gertler).[17]

Finally, taking stock of the literature on the supply of education in developing countries, Glewwe (with Michael Kremer) criticizes that, although school enrollment rates have risen rapidly in the developing world between 1960 and 2000, dropout rates remain high and learning outcomes disappointing, and thus argues that the primary policy question should be which policies most effectively improve learning, with RCTs as the preferred tool to conduct that investigation.[18] More recently, Glewwe has emphasized (with Karthik Muralidharan) that educational spending in developing countries could be much more cost effective, as improvements to pedagogy (e.g. remedial classes for children lagging behind) as well as improvements to school governance and teacher accountability tend to be much more cost effective than mere (yet widespread) increases in "standard" school inputs (e.g. more books); by contrast, interventions aimed at increasing the demand for education by raising students' returns to (or decreasing households' costs of) school enrollment and effort are also generally effective in improving learning outcomes, but vary widely in terms of cost effectiveness.[19]