The inverse productivity-size relationship has been observed the world over, yet there has not been an adequate explanation to describe why it occurs. Three hypotheses have been presented: imperfect markets, soil characteristics, and measurement error. This thesis focuses on soil quality and imperfect markets as the cause of the relationship: smaller plots are simply more fertile and thus more productive than larger, more marginal lands. Since soil characteristics have not been considered in previous analyses, the resulting omitted variable bias induces the inverse relationship. The soil hypothesis is tested within the context of imperfect markets, including various types of labor and household or village fixed effects. Using soil and labor data from Madagascar, this thesis finds that soil characteristics do not influence the inverse relationship The inverse relationship remains even after controlling for house-hold level effects questioning imperfect markets as a cause. Further work is warranted to ferret out the cause of the observed inverse productivity-size relationship. It may be allocative inefficiency of the household resources among the plots that gives rise to the inverse relationship.