We present a model in which investment in schooling generates two kinds of returns: the labor-market return, resulting from higher wages, and a marriage-market return, defined as the impact of schooling on the marital surplus share one can extract. Men and women may have different incentives to invest in schooling because of different market wages or household roles. This asymmetry can yield a mixed equilibrium with some educated individuals marrying uneducated spouses. When the labor-market return to schooling rises, home production demands less time, and the traditional spousal labor division norms weaken, more women may invest in schooling than men.

This paper provides an exhaustive characterization of testability and identifiability issues in the collective framework in the absence of price variation; it thus provides a theoretical underpinning for a number of empirical works that have been developed recently. We first provide a simple and general test of the Pareto-efficiency hypothesis, which is consistent with all possible assumptions on the private or public nature of goods, all possible consumption externalities between household members, and all types of interdependent individual preferences and domestic production technology. The test is proved to be necessary and sufficient. We then provide conditions for the identification of the sharing rule and the Engel curves of individual household members for a variety of different observational schemes.

Identifying Preferences under Risk from Discrete Choices by Pierre-Andre Chiappori, Amit Gandhi, Bernard Salanie and Francois Salanie. Published in volume 99, issue 2, pages 356-62 of American Economic Review, May 2009