This paper sheds light on the impact that conditional public cash transfers exert on living arrangement decisions by analyzing the effect of PROGRESA on the composition of beneficiary households. We use data collected for its evaluation and compare eligible households in communities receiving the Program to eligible households in communities not incorporated. Results based on double differences using propensity score matching compare the change in demographic composition, over a five year period, before and after the Program’s implementation. Results show that households that have benefited from the public aid are more prone to shelter new members of the extended family, who resided elsewhere prior to the intervention. Absent sons and daughters –alone or with their couples and young children-- are more prone to return home, while parents and grandparents of the household head, tend to adhere as well. We also find a partition effect of beneficiary households in which conditional transfers provide greater independence to young adults, sons and daughters of the household head, who detach from their original household and constitute their own families, even outside their community boundaries.

Recent research on household behavior suggests that, ceteris paribus, a woman’s "power" within a household influences consumption and time allocation choices. From an empirical point of view, a central stumbling block in this line of inquiry has been identification of sources of "power" that can plausibly be treated as exogenous. Aid to Families with Dependent Children, AFDC, was paid only to single women with children. The benefit level provides a natural fall-back for a low-income woman with children who is contemplating separation from her partner. As AFDC payments increase, separation will become more attractive and, we conjecture, the relative bargaining power of the woman in a household should also increase. If this is true, and if bargaining power does affect allocation decisions within the household, then the AFDC benefit level should affect household choices in intact families. This hypothesis is tested using the PSID from 1968 through 1992. Benefit levels, which (conditional on family size) vary across states and over time are treated as exogenous. In order to sweep out household-specific unobserved heterogeneity, models include household fixed effects. In addition, the model predicts the behavior of households with young children should be influenced by AFDC but not that of households with no children. Second, AFDC is unlikely to be paid to women in higher income households and so it should have a bigger influence on the behavior of lower income households. The results are consistent with these predictions. AFDC generosity does affect the allocation of resources in households with young children, and particularly lower income households with very young children. Corroborating evidence is drawn from the Consumer Expenditure Survey. We conclude that options outside marriage, as indicated by the generosity of AFDC benefits, affect bargaining power of women within marriage which, in turn, influences household resource allocation decisions.

The literature suggests men and women have different preferences. This paper exploits a random assignment social experiment in which women in treatment households were given a large public cash transfer (PROGRESA) and women in control households were given nothing. In an effort to disentangle the effect of additional income in the household from the effect of changing the distribution of income within the household, the impact of PROGRESA income is compared with all other income sources. In addition to spending on food, clothing and education, savings and investment decisions are examined. Additional money in the hands of women is spent on child goods (particularly clothing), improved nutrition (better quality diets) and investments in small livestock (which are traditionally managed and cared for by women). Among single headed households, PROGRESA income is not treated differently from other income. We conclude that not only do preferences of men and women differ but, among poor, rural Mexican households, resources under the control of women are more likely to be spent on investments for the future than resources that men control.