Tuesday, September 10, 2019

Universal Basic Income--Combined With What Else?

The idea of  a "universal basic income" has some immediate attraction. If we can land an astronaut on the moon, etc., etc. But along with other slogans like a guaranteed government job or single payer health insurance, the devil is in the details.

Three recent essays offer a useful overview of the choices and tradeoffs. Melissa S. Kearney and Magne Mogstad have written "Universal Basic Income (UBI) as a Policy Response to Current Challenges" for the Aspen Institute Economic Strategy Group (August 23, 2019). Also, the most recent issue of the Annual Review of Economics, published in August 2019, has a three-paper symposium on the universal basic income.
For those who don't have access to the Annual Review of Economics, the first paper is freely available here, the second paper is available as an NBER working paper here, and the third paper is freely available here.

An underlying theme of these discussions is that it is essentially meaningless to discuss the idea of a universal basic income in isolation. Whether you are talking about cost, or effects on distribution of income, or effects on work incentives, it matters considerably whether the universal basic income is views as an addition to any existing income transfer programs, or as a replacement for at least some of those programs.

For example, consider the question of cost.  Hoynes and Rothstein explain this way:
A universal payment of $12,000 per year to each adult U.S. resident over age 18 would cost roughly $3 trillion per year. This is about 75 percent of current total federal expenditures, including all on- and off-budget items, in 2017. (If those over 65 were excluded, the cost would fall by about one-fifth.) Thus, implementing this UBI without cuts to other programs would require nearly doubling federal tax revenue; even eliminating all existing transfer programs – about half of federal expenditures – would make only a dent in the cost. ...
A truly universal UBI would be enormously expensive. The kinds of UBIs often discussed would cost nearly double current total spending on the “big three” programs (Social Security, Medicare, and Medicaid). Moreover, each of these programs would likely be necessary even if a UBI were in place, as each addresses needs that would not be well served by a uniform cash transfer. Expenditures on other existing programs sum up to only a small fraction of the cost of a meaningful UBI. This suggests that a full-scale UBI would require substantial increases in government revenue. The impacts of whatever taxes are imposed to generate this revenue are likely of first-order importance in evaluating the impact of a UBI.
This insight helps to explain why no high-income country has actually adopted a "universal" basic income, and why most proposals for a "universal" basic income aren't really a simple universal payment. Instead, such proposals often include various phaseouts of the payments as other income rises, or rules that some of the money must be spent on purchasing health insurance, and so on and so forth.

On the issue of how a universal basic income would affect the distribution of income, the answer again depends on the extent to which is might replace other programs. The United States, like many other countries, uses "tagging" in its transfer programs, which means that transfer payments are often linked to some characteristic other than income. For example, payments may be linked to age (like Social Security), or to disability, or to whether or how many children are in a household (like Medicaid, the earned income tax credit, food stamps, and others).

Consider the proposal that is sometimes made for taking all the funds now spent on income transfers, and instead using that money for cash payments in the form of a universal basic income. (In "Universal Basic Income: A Thought Experiment" (July 29, 2014), I discuss one proposal along these lines for the US.)  As Hoynes and Rothstein explain, even if you cannibalized all spending on Social Security, Medicare, Medicaid, and every other program that involves government transfers, it wouldn't be enough to support a universal basic income of $12,000 per person. But set aside the cost arguments and instead focus on how the redistribution of income would be affected by moving away from a "tagging" system.

Hoynes and Rothstein do various calculations of how a universal basic income that replaces other government programs would affect who receives the funds. It shouldn't be any surprise that if you stop targeting the elderly, the disabled, and families with children, then households with those characteristics will get less. In contrast, households that are nonelderly, nondisabled, and with no children get tent to get more. They write:
This implies that were we to eliminate current income support programs and apply the funds towards a pure UBI, there would be a relative redistribution from low-earners to zero earners, but the first-order effects would be a massive distribution up the earnings distribution, along with a redistribution from the elderly and disabled towards those who are neither, primarily but not exclusively those without children.
As Kearney and Mogstad write: "The complexity of existing redistribution problems is a real issue, but the complexity is in large part based on seeking to address specific needs for specific groups: health care, housing, food, energy costs, and so on." For those attracted by the simplicity of just paying a flat cash amount to everyone, not linking benefits to family status or type of service, it's important to think seriously about what this shift away from tagging would be giving up.

Another main set of arguments about a universal basic income involves its interaction with labor markets. There are several arguments here that do not necessarily dovetail very well with each other. For example, some supporters of a universal basic income suggest that it will be needed in the future after the robot apocalypse makes most of human labor obsolete. When this actually happens, I'm ready to revisit this argument. But at present, the unemployment rate has been 4% or less for more tha a year and there are plenty of previous warnings about technological change would lead to permanent mass unemployment (here are examples from 1927, 1964, and 1982) that did not come to pass.

A gentler version of this argument is that a universal basic income would be a way of helping low-wage or low-income workers. But if helping a specific group of low-wage, low-income workers is the goal, then a "universal" payment is a peculiar way of accomplishing it. Instead, it would seem like an expansion of support for low-wage workers, perhaps designed in a way that is linked to work and provides an additional incentive to work, would make more sense.

Would a universal basic income discourage work? The direct evidence on this point remains thin. There have been studies of a few programs that make universal payments to certain groups, like the Alaska Permanent Fund (based on oil revenues from Alaska) or the Eastern Cherokee Native American tribe payments from gaming revenues, but the size of these payments is too small to be a stand-alone income. There have been experiments with something close to a universal basic income in Finland and Ontario, but these experiments were cancelled after a couple of years. There is some evidence from lottery winners, or from increases in disability insurance payments.

 Kearney and Mogstad provide an overview of the available evidence and argue that both economic theory and the existing evidence suggest that a true universal basic income--that is, an income received without any linkage to other income received, will tend to reduce work. In contrast, programs like wage subsidies, job training, or job subsidies seem likely to increase work. They write (citations omitted):
Studies of transfers that are more comparable in size to the types of UBI payments being proposed imply more negative labor supply effects. For example, a study of lottery winners find that, with an average annual prize of $26,000, each $100 in additional earnings reduced labor market earning by $11. A more recent study of lottery winners in Sweden also provides evidence of reduced earnings in response to winning a lottery prize. This study finds that winning a lottery prize leads to an immediate and persistent reduction in earnings. In addition, the effects of any guaranteed income program are likely to most strongly affect those marginally attached to the labor force. On this point, the lessons from expanded access to disability insurance payments is potentially instructive. Economists have found that the marginal beneficiary of a disability insurance award would have been almost 30 percentage points more likely to work had they not received benefits.
An intriguing thought that emerges from several of these papers is that the arguments for a universal basic income may be stronger for low-income countries. As Ghatak and Maniquet emphasize, most low income countries share several characteristics. A larger share of the population is close to subsistence, compared to high-income countries. As a result, a universal payment can help to raise a larger share of population out of poverty, and at a relatively low cost. In addition, the governments of low-income countries often have a hard time implementing detailed tax and welfare policies; for example, such governments may not be able to observe income levels or hours worked very accurately. Thus, linking government payments to income, as well as to disability, number of children, and even age may be more difficult. As they write, "UBI might be more appropriate in developing countries, especially those in which UBI could help circumvent the imperfections of government institutions in charge of helping the poor."

Of the papers I've mentioned here, Ghatak and Maniquet is the only one with a hefty share of math, and thus is likely to be a hard read for the unintiated. However, Banerjee, Niehaus, and Suri dig into the issues of a universal basic income for lower-income countries in more detail. They point out that while we don't have good evidence on pure universal basic income programs in low-income countries (although experiments are underway in some countries), we do  have a lot of evidence on programs in low-income countries that pay cash to recipients under various conditions. They write (citations omitted):
With the most current available data as of 2018, the World Bank identified 552M people living in the developing world who receive some form of cash transfer from their government. While none of these schemes were (to our knowledge) labelled as UBI, they all shared the common and crucial feature that recipients were given the freedom to do what they want with their money. Many transfers (particularly in South and Central America) were paid out conditional on certain conditions being met, but many others (particularly in Africa) were not. And in some cases - pensions, for example - these transfers have a structure (size, frequency, and duration) quite similar to UBI payments, though they are not universal.
What have we learned from the evaluation of these schemes? ...
First, evaluations generally have not found the negative impacts that many feared. Reviewing evidence on “temptation goods,” Evans and Popova (2017) find that transfers had on average reduced expenditure on temptation goods by 0.18 standard deviations. In other words, far from blowing their transfers on alcohol and tobacco, recipients appear to drink and smoke less. This finding in no way diminishes the seriousness of substance abuse as an issue for the poor, but it does suggest that lack of money may be a cause of substance abuse rather than a constraint on it. Turning to “dependency” ...  Banerjee et al. (2017b) find no systematic evidence that transfers discourage work.
Second, evaluations have found a great diversity of positive impacts. To give some sense, a partial list of outcomes affected in a positive way in one study or another ... includes income, assets, savings, borrowing, total expenditure, food expenditure, dietary diversity, school attendance, test scores, cognitive development, use of health facilities, labor force participation, child labor migration, domestic violence, women’s empowerment, marriage, fertility, and use of contraception, among others. ...
This variety implies that recipients value the flexibility that cash transfers provide: they reveal a preference for many different things. It also implies that a UBI is unlikely to appeal to a technocrat seeking cost-effective ways to increase any particular, narrow outcome.
In addition, they point out that a universal basic income may help economic growth in low-income countries by making it possible for low-income people to deal with the day-to-day risks they face and to make modest investments in small-scale entrepreneurship. And a universal benefit might build political support for a rudimentary social safety net in countries that do not yet have one.

On the other side, even if it is harder for a low-income country to run a precisely targeted income transfer program, imperfect targeting can still be useful. Rema Hanna and Benjamin A. Olken make this case in "Universal Basic Incomes versus Targeted Transfers: Anti-Poverty Programs in Developing Countries," in the Fall 2018 issue of the Journal of Economic Perspectives. They point out that a number of emerging-market countries make transfer payments conditional on behaviors, like whether children attend school or doctor visits, or else on observable characteristics like whether a home has a dirt floor, or a certain kind of roof or appliances. In some places, a "universal" payment requires taking enough time to register for the program or to undertake some work effort that those with higher income see no benefit from applying. In a few places, transfer payments are given to a community, which then must have a formal and open process for distributing those payments among the members of the community. They argue that a truly universal program, with payments going to people of all income levels, is less effective at addressing inequality than a program with imperfect targeting of benefit payments.

Here's a final comparison between universal basic income in high-income and low-income countries that struck as interesting, from the Banerjee, Niehaus, and Suri paper. They point out that one of the arguments for a universal basic income in low-income countries is that employment in such countries is often sporadic, with many people working a variety of part-time gigs rather than a single steady job, which is part of what makes it hard for the government in a low-income country to adjust benefits in response to income and work status. But of course, there is also concern that a greater share of workers in high-income countries are ending up in the "gig economy"  with a series of part-time jobs, which in turn makes it more challenging for high-income countries to set up programs where transfer programs will make sporadic payments in the gaps between sporadic jobs. Banerjee, Niehaus, and Suri write:  
"[D]eveloping countries already look like one possible future for the developed ones: few people hold stable full-time jobs, many work a variety of part-time gigs instead, and as a result, public policy has never been based on an assumption of universal full-time employment. Perhaps in this there is something the rich countries can learn from the poor."