For as long as there have been government programs designed to help the poor, there have been critics insisting that helping the poor will keep them from working. But the evidence for this proposition has always been rather weak.
And a recent study from MIT and Harvard economists makes the case even weaker. Abhijit Banerjee, Rema Hanna, Gabriel Kreindler, and Benjamin Olken reanalyzed data from seven randomized experiments evaluating cash programs in poor countries and found "no systematic evidence that cash transfer programs discourage work." Attacking welfare recipients as lazy is easy rhetoric, but when you actually test the proposition scientifically, it doesn't hold up.
No effects in Honduras, no effects in Indonesia, no effects in Morocco…
The programs covered in the study have a pretty wide geographic spread. There are four in Latin America (two in Mexico, one each in Nicaragua and Honduras), two in Southeast Asia (Philippines and Indonesia), and one in Morocco.
Most of the programs the study analyzes are what's known as "conditional cash transfers" (CCTs), where households receive help on the condition that they, say, have their kids attend school, or get them vaccinated. The idea is both to help poor people and to use the aid as a lever with which to ensure kids are getting educated and receiving health services. CCTs first caught on in Latin America, so it makes sense that most of the programs analyzed in the paper are from countries in that region. But the study also includes a Mexican program that provided a $13-a-month unconditional cash transfer to families in poor regions.
Exactly zero of the seven programs saw a statistically significant change in either employment levels or hours worked per week:
In some trials, work went up; in others, work went down. In none of them was the change substantial. For every program except those in Honduras and the Philippines, the data was comparable enough that the researchers could pool it and estimate effects across the programs. That allows more precise estimates than you could get from any of the five comparable studies alone. The 95 percent confidence interval for how the programs affected the employment rate ranged from a 1.6 percentage point decline to a 0.9 point increase. There just isn't any change happening here.
Some other studies find that cash encourages work
To some degree, this actually undersells cash programs. Two recent RCTs have suggested that giving cash to poor people in the developing world could actually, in some cases, encourage work. One paper by Christopher Blattman, Nathan Fiala, and Sebastian Martinez evaluated a program that gave cash grants ($382 per person, on average) to groups of young, unemployed Ugandans to help them learn skilled trades, and found that hours of work rose by 17 percent, and earnings by 38 percent. Another paper, by Blattman, Eric Green, Julian Jamison, and Jeannie Annan, looked at a program in Nigeria that gave about $150 and some basic business skills training to women in northern Uganda. Work hours increased by 61 percent.
These programs are importantly different from the ones evaluated in the Harvard/MIT paper. For one thing, they were oriented around supporting business activities: The program for women provided training and supervision, and the program for young adults required applicants to prepare proposals for how they'd use the cash to further their business. They also involved a one-time grant, rather than an ongoing transfer, and generally operated at a smaller scale than the government programs Banerjee et al. evaluated.
But the reasoning that makes welfare opponents skeptical about cash transfer programs should also make them worry about giving poor people big lumps of cash to use as business capital; why wouldn't they just take the cash and spend it? So the success of these programs provides yet another counterargument to welfare queen paranoia.
It's not just abroad
All of the above evidence concerns the developing world. But it's worth being skeptical about welfare queen claims in rich countries as well. For one thing, the biggest program the US currently runs for prime-age poor adults is the earned income tax credit. There's a substantial body of evidence showing that the EITC encourages work, usually by pulling single parents into the workforce. That lets it have an anti-poverty impact beyond the actual cash that the tax credit provides to families.
But even unrestricted cash programs aren't likely to have a major effect on work in rich countries. A number of studies in the US in the 1970s examined "negative income tax" programs, where a set sample of poor households received cash grants whose size shrunk as the households earned more money through their jobs. The studies found very mild declines in work, largely due to people taking longer to find a good job while unemployed and spending longer in school. Even those estimates were exaggerated by participants underreporting the amount of work they were doing, perhaps to get bigger checks; when researchers examined administrative data, rather than survey responses, they found barely any effect on work at all.
A much better experiment in Canada, where an entire town got a guaranteed income by way of a negative income tax, found even milder reductions in work, and then only with new mothers (who spent longer at home with their newborns) and teenagers.
There's no doubt that poorly designed social programs can deter work. Aid to Families With Dependent Children, the pre–welfare reform welfare program, was found to decrease hours worked by 10 to 50 percent among recipients; that likely has something to do with the fact that AFDC benefits were taken away at a rate of 100 percent, so every dollar earned on the job was a dollar not received from AFDC. Who would work under that condition?
But most welfare programs are better than AFDC. Whether they're in the US or in developing countries, they don't tend to keep people from working.