I analyze the impact of the early 1990s state waivers from Aid to Families with Dependent Children (AFDC) guidelines . . . . AFDC waivers decreased the public assistance available to impoverished divorced women and thereby reduced a woman's bargaining threat point in marriage. . . .
[D]ecreases in potential welfare benefits altered the expenditure patterns of two-parent families. Waivers were associated with increased expenditure on food at home relative to restaurant meals and decreased expenditure on child care and women's clothing, suggesting greater home production and decreased consumption by women. Such changes are evident only for households containing a woman with a reasonable probability of receiving welfare benefits if her marriage ended.
While the theory behind why welfare benefits should influence how women behave within marriage is solid, I have doubts about whether the magnitude of the impact of a tweak in the AFDC program is great enough to be measurable, as opposed to being a correlation but not a cause of the relationship between the AFDC waivers and the behavior. I am surprised that the decision to end a marriage would be so finely balanced that this kind of factor could have a meaningful influence. Put another way, I'd expect that other factors, like differences in local divorce laws or differences in local cultural norms about marriage, would swamp this effect.
On the other hand, given the fragility of modest income families in the United States today, perhaps it shouldn't be so surprising. Economic dependency is pretty modest in this income bracket relative to typical upper middle class couples, and so other factors may be more important.