Mothers’ Pensions and Female Headship

Longitudinal studies of white female headship over the past few decades do not find significant welfare effects but do find a positive relationship between state fixed effects and welfare benefit levels. In other words, although white female headship does not respond to the year-to-year changes in welfare benefit levels, states with the most generous welfare benefits tend to be the states with the highest rates of white female headship. This positive relationship does not exist, however, for blacks. The questions are: Why does a positive relationship exist for whites and when did it originate? and Why does this relationship not exist for blacks? This paper examines these questions by considering the relationship between female headship for blacks and whites and state mothers’ pensions legislation enacted between 1910 and 1920. Mothers’ pensions programs were the first public cash assistance programs targeted to single mothers and, like their successors, varied greatly across the states. The results indicate that welfare generosity preceded relatively high rates of white female headship. The positive relationship between welfare generosity and white female headship observed today was not embodied in the mothers’ pensions legislation enacted in the 1910s. But these early welfare programs, like more recent programs, were not responsive to, and may in fact have been reactionary to, the experiences of black women. The author thanks Steven Nafziger and Jennifer Nou for able research assistance and seminar participants at the NBER/DAE Summer Institute 2001 and the University of Illinois for helpful comments and suggestions. In 1970, 12 percent of families with children under the age of 18 were headed by single mothers. By 2000, that fraction had increased to 26 percent (U.S. Census Bureau, 2001). This dramatic change in American family structure has been cause for some concern. Families headed by single mothers have high rates of poverty, and children raised in such families are more likely to drop out of school, have children out-of-wedlock, and have difficulties in the labor market in their young-adult years (see McLanahan and Sandefur 1994). In the search to explain the increase in single-mother families in the past few decades, much of the attention has been focused on the relationship between female headship and the American welfare system. In most states, only families headed by single parents are eligible for cash assistance programs. Accordingly, some have argued that the welfare system encourages divorce and separation and discourages marriage. Economists have attempted to examine this issue by estimating the relationship between female headship and the level of welfare benefits. Early analysis used cross-sectional data and relied on the variation in benefit levels across the states to identify the “welfare effects.” These studies tended to find statistically significant welfare effects: as benefits increased, women were less likely to marry and more likely to head households. This evidence would seem to support the view that the welfare system discourages the formation of two-parent families. When the analysis was extended to repeated-crosssection and longitudinal data, however, a slightly different and more complicated story emerged. Moffitt (1994) and Hoynes (1997) found that controlling for state fixed effects causes the relationship between benefit levels and female headship for whites to disappear. As both authors document, this change occurs because state fixed effects are positively correlated with state benefit levels. In other words, states that offer high levels of benefits are those with high rates of female headship, and states that offer low levels of benefits are those with low rates of female headship. The disappearance of the benefit level effect when state fixed effects are included in the model indicates that white female headship does not respond to the year-to-year changes in the level of benefits. But the basic conclusion of these models is still that 1 See Moffitt (1992) for a review of this literature. See also Schultz (1994).