with Eva Vivalt, Andrew Goodman-Bacon, Elizabeth Rhodes, and Alex Bartik, revise and resubmit, Journal of Economic Literature
Abstract The early 2020s saw a wave of experimentation with guaranteed basic income, or GBI, which provides repeated unconditional cash transfers. In this article, we synthesize evidence from 34 randomized controlled trials of GBI in high-income countries. Collectively, these studies enrolled 54,793 participants; however, high rates of attrition in many studies meant the effective collective sample size was much lower. Cash transfer sizes were large: nearly 60\% of baseline income in terms of present discounted value. Using Bayesian hierarchical meta-analysis, we examine effects on employment, household income, food insecurity, mental distress, and the ability to meet emergency expenses. Transfers modestly reduced food insecurity and improved participants’ ability to cover emergency expenses but produced little improvement in mental distress. Average employment effects were near zero, although studies with lower differential attrition indicated a two-percentage-point decline, accompanied by small reductions in household income. To place these results in context, we compare the estimated labor-supply responses with evidence from the Negative Income Tax experiments, lottery winners, and other quasi-experimental studies of unearned income. After scaling interventions by the present discounted value of transfers, recent guaranteed-income experiments imply larger reductions in employment and earnings than these earlier studies, particularly in the lower-attrition subsample. Nevertheless, the overall effects of guaranteed income were modest relative to both advocates’ hopes and critics’ concerns. High and differential survey attrition, small samples, inconsistent outcome definitions, and limited variation in transfer amounts and program duration constrain inference. Future experiments should prioritize coordinated measurement, administrative data linkage, sustained follow-up, and greater variation in program design.
with Elizabeth Rhodes, Alex Bartik, David Broockman, Patrick Krause and Eva Vivalt, “reject” and resubmit, American Economic Review
Abstract This paper examines the impact of a large, randomized cash transfer on parental behaviors, investment in children, children's social, behavioral, and educational outcomes, and pregnancy and childbearing. We find that parents who were randomly selected to receive a $1,000 per month unconditional cash transfer for three years spent more on their children each month and reported better parenting behaviors (such as supervising their children more closely) compared to those randomized to receive $50 per month over the same period. However, possibly due to this closer monitoring, parents in the treatment group also reported that their child was experiencing more developmental difficulties and stress. Parents with the lowest incomes at baseline experienced the largest improvements in parenting; among these parents, the transfer also increased the use and quality of non-parental child care. The transfer did not have a meaningful effect on most educational outcomes measured in school administrative records, nor did it affect characteristics of the home environment, child food security, exposure to homelessness, or parental satisfaction. Although treated families were more likely to move, we did not detect changes in most measures of neighborhood quality, though proximity to child-focused amenities such as daycares appeared to increase in the treatment group relative to the control group. The transfer did not affect childbearing, pregnancy, or outcomes related to contraception. While the transfer reduced parents' stress and mental distress in the first year of the program, these effects were short-lived and dissipated by the second year of the transfer, analogous to what was documented previously in the full population of participants.
with Elizabeth Rhodes, Alex Bartik, David Broockman, Karina Dotson, Patrick Krause and Eva Vivalt, NBER Working Paper
Abstract We study the causal effects of income on political attitudes and behavior with a field experiment. In the experiment, a non-profit gifted 1,000 low-income Americans $1,000 per month for three years tax-free, and 2,000 control participants $50 monthly. Contrary to resource models of participation, we find no effects on political participation or engagement, and rule out effects equivalent to the observational association between turnout and income. Political preferences largely do not change, with the estimates again distinguishable from the observational relationship that economic conservatism increases with income. Dispositions such as trust in government, polarization, and support for democracy also do not change. We do find effects consistent with mood misattribution: affect towards one's own racial group, other racial groups, and some politicians slightly improves. There is also some evidence that treated participants saw work as more important for individuals, society, or even as a requirement for accessing government programs; qualitative evidence illuminates potential mechanisms. Our findings contrast with findings from other economic shocks such as government-sponsored or taxable transfers—thereby helping clarify the mechanisms likely responsible for their effects—and underscore the durability of political predispositions.
with Elizabeth Rhodes, Alex Bartik, David Broockman, Patrick Krause and Eva Vivalt, NBER Working Paper
Abstract We provide new evidence on the causal effect of unearned income on consumption, balance sheets, and financial outcomes by exploiting an experiment that randomly assigned 1000 individuals to receive $1000 per month and 2000 individuals to receive $50 per month for three years. The transfer increased measured household expenditures by at least $300 per month. The spending impact is positive in most categories, and is largest for housing, food, and car expenses. The treatment increases housing unit and neighborhood mobility. We find noisily estimated modest positive effects on asset values, driven by financial assets, but these gains are offset by higher debt, resulting in a near-zero effect on net worth. The transfer increased self-reported financial health and credit scores but did not affect credit limits, delinquencies, utilization, bankruptcies, or foreclosures. Adjusting for underreporting, we estimate marginal propensities to consume non-durables between 0.44 and 0.55, durables and semi-durables between 0.21 and 0.26, and marginal propensities to de-lever of near zero. These results suggest that large temporary transfers increase short-term consumption and improve financial health but may not cause persistent improvements in the financial position of young, low-income households.
with Elizabeth Rhodes, Alex Bartik, David Broockman, Patrick Krause and Eva Vivalt, conditionally accepted, American Economic Review
Abstract We study a randomized trial in which 1,000 low-income U.S. adults age 21 to 40 received \$1,000 per month for three years, from November 2020 until October 2023, while 2,000 control participants received \$50 monthly over that same period. The transfer produced large but short-lived improvements in stress and food security and increased hospital, emergency department, and dental care use. However, it did not improve self-reported physical health, clinical biomarkers, or mental health after the first year, and we can rule out small effects on many outcomes. We also find no effects on the health of children in participants' households.
with Laura R. Wherry, revise and resubmit, AEJ: Economic Policy
Abstract Undocumented immigrants are ineligible for public insurance coverage for prenatal care in most states, despite their children representing a large fraction of births and having U.S. citizenship. In this paper, we examine a policy that expanded Medicaid pregnancy coverage to undocumented immigrants. Using a novel dataset that links California birth records to Census surveys, we identify siblings born to immigrant mothers before and after the policy. Implementing a mothers' fixed effects design, we find that the policy increased coverage for and use of prenatal care among pregnant immigrant women, and increased average gestation length and birth weight among their children.
