How “Effective” Is Guaranteed Income?

Not as much as advocates amplified by the New York Times say

On the front of the business section of Saturday’s New York Times, an article by David Streitfeld touted “the idea of a guaranteed income — also known as cash transfers, unconditional cash and, in its most utopian form, universal basic income.” 

“The data seems to show that the programs are effective,” the Times news article said. “Critics wondered if recipients would blow the funds on lottery tickets and booze. Dozens of pilot programs that took fewer years than the Unconditional Income Study have since answered that question.” 

“While they welcomed Mr. [Sam] Altman’s study, the issue for members of the basic income community has shifted to establishing the programs on a wider scale. The time for research, they say, is over,” the Times reported.

It’s an overstatement to say “the data seems to show that the programs are effective.” The available data is quite limited. It largely consists of surveying people who got the extra cash payments and having them self-report that, with the extra money, they had an easier time making ends meet. That’s almost tautological. These surveys are hard to conduct, because people poor enough to qualify for the extra cash frequently drop out of either the studies or the control group. A lot of the pilot programs took place during the Covid pandemic and during the government spending binge on Covid relief, making it hard to disentangle the effects of the pilot basic income programs from other government spending programs. The test programs were also small, and short-term.

For example, a Cambridge, Massachusetts program provided monthly $500 payments for 18 months to 130 households, and compared those households to a control group of 156 others. According to Appendix B of a study of the program by the University of Pennsylvania’s Center for Guaranteed Income Research, by six months into the study, half of the overall participants (recipients and control group members together) had dropped out. The Cambridge study itself also acknowledges, “the unique socio-economic conditions of the pandemic era could have significant bearing on the study’s findings, affecting their applicability and relevance to other times and settings.”

A Stockton, California program ran for 24 months and included 125 recipients alongside a 200-person control group, a size the study authors themselves concede is “small.” That survey, also by a team from the University of Pennsylvania’s Center for Guaranteed Income Research, also suffered from huge attrition; “retention was approximately 35% in control and 55% in treatment,” which is a fancy way of saying that 65 percent of those in the control group and 45 percent of those in the treatment group dropped out of the study before it was completed. The people kept getting the money even if they didn’t fill out the surveys; “A key tenant [sic] of guaranteed income is unconditionality; thus, even members of the treatment group were not compelled to participate in research activities as a condition of receiving the guaranteed income.” 

On Monday, after the Times article appeared, two National Bureau of Economic Research working papers came out, based on a larger survey—a 1,000-person treatment group and a 2,000-person control, split between Illinois and Texas.

One paper, “Does Income Affect Health? Evidence from a Randomized Controlled Trial of Guaranteed Income,” by a five-person team including the University of Michigan’s Sarah Miller, found, “The results from this experiment suggest that, for the sample and time frame we study, directly reducing poverty via cash transfers was not effective at improving health outcomes. We document a precise null effect for a variety of physical health outcomes and our point estimates are able to rule out even very small improvements in physical health. Improvements we observe in mental health are short-lived and, by the second year of the transfers, the treatment group reports no better mental health than control group members. Furthermore, the intervention did not appear to improve access to medical care, nor did it lead to participants making other health investments via sleep and exercise.”

A second paper, “The Employment Effects of Guaranteed Income: Experimental Evidence from Two U.S. States,” by another team that included Miller, found “The transfer caused total individual income to fall by about $1,500/year relative to the control group, excluding the transfers. The program resulted in a 2.0 percentage point decrease in labor market participation for participants and a 1.3-1.4 hour per week reduction in labor hours, with participants’ partners reducing their hours worked by a comparable amount. The transfer generated the largest increases in time spent on leisure, as well as smaller increases in time spent in other activities such as transportation and finances. …we find no impact on quality of employment, and our confidence intervals can rule out even small improvements. We observe no significant effects on investments in human capital.” Give people free money, and they and their spouses work less than they did before.

If “the data seems to show that the programs are effective,” was an overstatement on Saturday, then today, after the release of these two new studies, it’s downright inaccurate. 

You wonder why the Times ran a story with the claims that “the time for research is over,” rather than waiting a few more days and reporting on the research.

If philanthropists or state or local governments want to keep experimenting on this and tracking the results with rigorous, long-term research, fine. But for advocates of these programs to claim that “the time for research is over” is, at this point, really too much. It can take decades to fully understand the results of welfare programs like these, or to weigh the costs and benefits of sending people cash against the government spending money on something else, or against simply lowering taxes by a similar sum. That Silicon Valley AI guy Sam Altman takes a lot of flak, some of it probably deserved, but give him and his team credit for running and reporting a serious study on this policy proposal. With luck, the research findings may slow the rush to roll out these guaranteed income programs nationwide on the basis of “data” that amount to less than advocates, and the New York Times, would have the public believe.

Thank you: At The Editors, our income is tied to the number of paying readers. Thanks to those of you who are in that group. If you haven’t yet signed up, please do. Where else can you find the combination of press criticism, social science research design criticism, and market economics? If you sign up, we won’t blow the funds on lottery tickets and booze; we’ll use the money to fuel our continued growth and sustain our editorial independence.