Several studies find that child labor incidence is higher in households with larger land holdings. The existence of this “wealth paradox” has been explained as the consequence of simultaneous imperfections in the land and labor markets. This work shows that although rural households in Colombia and Mexico seem to exhibit this same positive relationship between land and child labor, the wealth paradox disappears when individuals are evaluated using longitudinal data. A possible explanation for this is that the omission of idiosyncratic household preferences regarding schooling, child labor and land holdings in cross-sectional estimates leads to an overestimation of the effect land has on these outcomes.