AVOIDING THE SNAKE PIT
There is no evidence that cash transfers make people work less.30 Economists find this surprising—why would you work if you did not need the money to survive? What about the temptation of sloth, for which the biblical punishment is being thrown in a snake pit in hell?
It seems plausible that many (perhaps most) people genuinely aspire to do something with their lives, but the exigencies of surviving on very little paralyze them.
Perhaps getting the extra cash encourages them to work harder and/or try new things. In Ghana, Abhijit and his colleagues carried out an experiment. Beneficiaries were offered a chance to make bags, which the experimenters then bought from them at very generous prices. Some of the women workers (chosen at random) were also part of a program that gave them a productive asset, most often goats, along with some training in how to make best use of the asset and some confidence boosting (these were very poor women, who did not necessarily believe they could be successful at anything). Despite the fact that tending the goats added to their workload (and also gave them some income, so that they were less desperately in need of extra cash), the women who got into the program produced more bags and earned more from them than those who were not included in the asset giveaway. Most interestingly, the big difference between those with and without an asset became evident when a bag had a complex design. Asset beneficiaries worked faster, yet met the necessary quality standards. The most plausible explanation is that getting the gift of the asset freed them up from worries of survival, giving them the bandwidth and the energy to focus on their work.31It is also true that the typical poor person in the developing world cannot get a loan (or can only get one at some astronomical interest rate) and has no one to bail them out if their venture fails.
Both these conditions make it much harder for them to start the business of their dreams. A cash transfer that goes on for some years both provides some extra finance and backstops consumption if the enterprise fails. Perhaps a guaranteed income would make the poor willing to go somewhere else to look for a better job, to learn a new skill, or to start a new business.But maybe all of this applies only in developing countries, where the poor are very poor indeed and the cash actually enables them to work. Perhaps in the United States things are very different since everyone, however poor, is usually able to find work. Is it possible that the sloth effect dominates there? As it turns out, there is also evidence, going back to the 1960s, suggesting that sloth should not be a major concern in the US. In fact, the first-ever large-scale randomized experiment in the social sciences, the New Jersey Income Maintenance Experiment, was devised precisely to determine the impact of the “negative income tax.” A negative income tax (NIT) implements the idea that the system of income taxation should be designed so everyone is guaranteed to receive at least a minimum income. The poor should pay negative taxes so they get paid more than they earn, but as they get richer they will get less in transfers until at some point they start paying into the system.
This is different from a UBI, because for people near the tipping point between being takers from and payers into the system, there is potentially a strong disincentive to work. In other words, in addition to the income effect (I do not need to work if I have enough money to survive on already) that most policy makers worry about, such schemes can have a substitution effect (working is less valuable since what I make in extra income is taken out as reduced welfare payments).
Many scholars and policy makers, in both political parties, were in favor of a negative income tax. On the left, the US Office of Economic Opportunity under Democrat president Lyndon Johnson heralded the idea and set forth a plan for replacing traditional welfare with an NIT.
On the right, Milton Friedman advocated replacing most existing transfer programs with a single NIT. Republican president Richard Nixon proposed it in 1971 as part of his package of welfare reform, but Congress did not approve it. A key concern at the time was that beneficiaries would work less as a result of the program, and thus the government would end up paying people who would otherwise have earned their own living.It was then that Heather Ross, a PhD student in economics at MIT, came up, arguably for the first time in economics, with the idea of an experiment to settle this issue. Ross was frustrated that politicians used anecdotes to justify economic policy, and that there was no factual basis to establish whether low-income people would stop working if they received help from this kind of program. In 1967, she submitted a proposal to the US Office of Economic Opportunity for an RCT. This was ultimately funded and, as Ross put it, she ended up with a “$5 million thesis.”32
The outcome of this inspired proposal would be not just a New Jersey NIT experiment, but also a series of others. In the early 1970s, Donald Rumsfeld (yes, the same one) steered the NIT away from full implementation and toward a series of experiments. The first experiment took place in urban areas in New Jersey and Pennsylvania (1968–1972), with subsequent experiments in rural areas of Iowa and North Carolina (1969–1973), in Gary, Indiana (1971–1974), and the largest one, the Seattle-Denver Income Maintenance Experiment (SIME/DIME) experiment, in Seattle, Washington, and Denver, Colorado (1971–1982, covering forty-eight hundred households).33
The NIT experiments convincingly established the feasibility and usefulness of running RCTs for policy making. It would be decades before comparably intellectually ambitious projects would take center stage again in social policy. That said, these being the first experiments in the social sciences, it is not surprising their design and implementation were far from perfect.
Participants were lost, samples were too small to get precise results, and, most worryingly, data collection was contaminated.34 Moreover, because the experiment was short lived and small scale, it was also not easy to extrapolate what would happen in response to a more permanent and more universal program.Nevertheless, taken together, the results suggest the NIT program did reduce labor supply a bit, but not nearly as much as was feared. On average, the reduction in time worked was only between two and four weeks of full-time employment over a year.35 In the largest experiment (SIME/DIME), husbands who received the NIT reduced their hours of work by only 9 percent compared to those who did not, although wives who received the NIT reduced their hours by 20 percent.36 Overall, the official conclusion of the study was that the income maintenance program did not have large effects on the propensity for people, particularly the prime earners in the family, to work.37
There are also examples of local unconditional transfer programs, from various parts of the United States. The Alaska Permanent Fund has distributed, since 1982, a yearly dividend of $2,000 per person per year. It seems to have no adverse impact on employment.38 Of course, the Alaska Permanent Fund, while universal and permanent (as its name indicates), is also quite small compared to the proposed UBI. If it had been sufficient to live off, people might have stopped working. A more UBI-like program is the payment of casino dividends on Cherokee lands to the members of the tribe. The transfers, about $4,000 per adult per year, represent a large boost in income, since the per capita household income for Native Americans is about $8,000. Comparing eligible and ineligible families in the Smoky Mountains before and after the payments, one study found no effect on work in the family, but large positive impacts on the education of the adolescents.39