A new working paper for the National Bureau of Economic Research points to significant long-term benefits for children in poverty whose families received cash transfers through the first U.S. government welfare program. Researchers used census, World War II and death records to compare male children of mothers who received help through the federal Mothers’ Pension program between 1911 and 1935 to male children whose mothers applied for help but were rejected from the program. You can read the full research paper here (pdf) or access it here. I’ve posted a few excerpts after the jump.

The main takeaway: “Male children of accepted mothers received one-third more years of schooling, were less likely to be underweight, and had higher income in adulthood than children of rejected mothers.” The study did not include female children of early welfare recipients because name changes through marriage make it much harder to track long-term outcomes for girls.  

Excerpt from THE LONG TERM IMPACT OF CASH TRANSFERS TO POOR FAMILIES by Anna Aizer, Shari Eli, Joseph Ferrie, and Adriana Lleras-Muney.

We ask whether targeted cash transfers improve children’s long-run outcomes, with a particular focus on longevity, by studying the Mothers’ Pension (MP) program-the first government sponsored welfare program for poor families with dependent children in the United States (1911-1935). The intent of the MP program was to improve the conditions of “young children that have become dependent through the loss or disability of the breadwinner” (Children’s Bureau, 1933). The transfers generally represented 12-25% of family income, and typically lasted for three years. In 1935, the MP program was replaced by the federal Aid to Dependent Children (ADC, later Aid to Families with Dependent Children, and now TANF).

One of the main challenges in evaluating whether cash transfers (or any public program) improve outcomes is identifying a plausible counterfactual: what would children’s lives have been like in the absence of receiving transfers? Our main strategy is to use as a comparison group the children of mothers who applied for transfers but were denied. This strategy of comparing accepted and rejected applicants for program evaluation has been used successfully in studies of disability insurance (von Watcher et al. 2011, Bound 1989). Its validity depends on the extent to which accepted and rejected mothers and their children differ on unobservable characteristics. We document that rejected mothers were on average slightly better-off based on observable characteristics at the time of application, consistent with the information in administrative records that applicants were most often rejected because they were deemed to have sufficient support. Under the assumption that accepted and rejected applicants are otherwise similar, the outcomes for boys of rejected mothers provide a best-case scenario (upper bound) for what could be expected of beneficiaries in the absence of transfers. Thus our estimates are, if anything, likely to understate the benefits of the program. We also construct alternative counterfactual groups based on differences in eligibility across counties to examine the robustness of our findings.

The second challenge is to obtain data containing long-term outcomes for a large sample of recipients and plausible comparison groups. […] Administrative individual records from the early years of the ADC program (1935-1962) have been lost or intentionally destroyed. Although records do exist for recipients of Aid to Families with Dependent Children (AFDC), the program that replaced ADC in 1962, these cohorts are too young for us to evaluate the impact on longevity.

Another important limitation of studying AFDC recipients is that they were also eligible for multiple in-kind transfers such as Medicaid, housing and food stamps, which makes it impossible to evaluate the impact of cash transfers alone.2

Instead, we use administrative records from the precursor to the ADC program-the Mothers’ Pension program (1911-1935). These data have the advantage of (1) being available in large numbers and containing both accepted and rejected applicants; (2) containing identifying information that allows us to link the children with other datasets to trace their outcomes; and (3) including children who were born sufficiently long ago that we can measure their longevity.3

Using data on 16,000 children from eleven states who were born between 1900 and 1925 and whose mothers applied to the Mother’s Pension program, we find that receiving cash transfers increased longevity by about one year. These effects are driven by the poorest families in the sample; for them, the longevity increases are larger (about 1.5 years of life). These results are very robust to alternative functional form specifications, alternative counterfactual comparisons (e.g. divorced women in some counties, but not others, were eligible for MP benefits), and our treatment of attrition. Because income transfers were the only major public benefit that poor children were eligible for until 1950 (with the exception of public schooling), we can interpret our results as the effect of cash transfers alone.

To investigate potential mechanisms behind the positive effect on longevity, we explore whether cash transfers in childhood are associated with improvements in education, income and nutritional status, since previous work has documented that all three have positive and independent effects on mortality. To do so, we match a subset of our records to WWII enlistment and 1940 census records. The results suggest that cash transfers reduced the probability of being underweight by half, increased educational attainment by 0.4 years, and increased income by 14 percent during adulthood (ages 20-45). […]

We conclude that cash transfers to poor families during the first part of the twentieth century ameliorated early life conditions enough to improve both medium- and long-term outcomes of boys growing up in poverty. While conditions today differ significantly from those at the beginning of the twentieth century, suggestion caution in drawing conclusions as to the anticipated impact of cash transfers in the twenty-first century, three important similarities remain and suggest the current relevance of the effects we document. First, both the MP program and current welfare programs target children in female-headed households. These children were, and continue to be, the poorest children. Second, historical comparisons presented in the last section of the paper reveal that income played an important role in producing positive child outcomes, both today and at the beginning of the twentieth century, when the MP program operated. Finally our short- and medium-term effects on education and health are consistent with contemporary evidence on the effect of poverty-reduction programs in the US and in developing countries. Altogether these results suggest that targeted cash transfers are also likely to improve lifetime outcomes today. Though we acknowledge our study’s limitations, the historical evidence we employ are the only means presently available to assess the impact of cash transfers across the entire life course.