Abstract

I use variation in coffee production to measure changes in local economic conditions, which proxy for the value of children's time. I test how this short-term variation affects child labor and schooling in Brazilian coffee regions using seven rounds of household surveys. Increases in the county-level value of coffee production led to more work among middle-income boys and girls, poorer children were withdrawn from school, while richer children were not affected. Thus, during periods of economic growth, education of the poor may be adversely affected. © 2006 Elsevier B.V. All rights reserved.