A Social Safety Net for an Age of Uncertainty
How to create a more accessible and usable experience one small decision at a time
The COVID-19 pandemic is turning out to be a wake-up call, not just for public health, but for economic security, as well. We are learning that any one of us can be knocked off the ladder of prosperity at any time, no matter what rung we are on today.
What is more, COVID-19 is by no means the last crisis we are likely to face. Just a few things that may be coming down the pike:
- A really bad virus, one as contagious as measles and as deadly as Ebola.
- Floods, wildfires, droughts, and famines brought on by climate change.
- A job apocalypse, in which entire middle-class occupations disappear one by one.
Meanwhile, none of the old risks are going away. The risk of being born poor, making even the first step to self-sufficiency precarious. The risk of being born with a disability or congenital illness. The risks of bad choices — dropping out of school; falling into crime; falling into substance abuse.
This crisis has shown everyone, left, right, or center, just how inadequate our fragmented social safety net is for dealing with what the world is throwing at us. Encouragingly, the first response has been a sound one: Send money, and send it fast. To think how radical the idea of universal cash assistance seemed, oh such a short time ago.
The idea that cash is what people need most when adversity strikes is far from new. Decades ago, Milton Friedman argued for programs that give help “in the form most useful to the individual, namely cash.”
More recently, Charles Kenny cites examples from around the world in support of a simple policy recipe: “Give poor people cash without conditions attached, and it turns out they use it to buy goods and services that improve their lives and increase their future earnings potential.”
However, the current push for emergency cash assistance is ad hoc and impulsive. So be it. Get the first checks in the mail fast. But meanwhile, we should give some careful attention to designing a more orderly system of cash assistance, custom-tailored for an age of uncertainty.
The requirements are clear: The new system should be seamless, work-friendly, and resilient. But what form should it take? A universal basic income? A negative income tax? Wage subsidies? This commentary examines each of these alternatives and concludes by recommending a reform program that including features of each of them.
A universal basic income
The simplest cash assistance policy is a universal basic income (UBI) that gives everyone an equal basic grant regardless of earnings or work status.
Although the “U” in UBI implies that the basic grant is available to everyone, in practice some broad standards of eligibility, such as age or citizenship status, would be needed to define who was meant by “everyone.”
For example, the “Freedom Dividend” that was the centerpiece of Andrew Yang’s campaign for the Democratic nomination limited eligibility to citizens of the United States over 18 years of age. A universal child allowance could be paired with an adult UBI, or a uniform basic grant could be offered to everyone regardless of age. Some UBI advocates would include at least some people who are residents, but not citizens.
Income security. The primary goal of cash assistance is income security. Income security is not just a matter of one’s average level of income, but also of its reliability and of the availability of reserves to meet emergencies. For the sake of discussion, I will use a benchmark based on an average income at least equal to the federal poverty level (FPL).
A UBI would be fully effective in providing income security, so defined, only if the basic grant itself were above the FPL for every recipient. (The FPL for an unrelated individual is $12,760 in 2020, and $26,200 for a family of four.) However, even a considerably smaller basic grant would improve on the present situation.
For example, suppose the basic grant for everyone were set at half the FPL for an unrelated individual, or $6,380 per year. The Census Bureau refers to half the FPL as deep poverty. A UBI with a basic grant of that amount would eliminate deep poverty for all unrelated individuals, and would fully eliminate official poverty for the roughly 50 percent of unrelated individuals who have income deficits less than half of the FPL.
Unrelated individuals account for only about 20 percent of the poverty population. For larger households, the FPL increases less than in proportion to the number of household members, so a UBI with a basic grant of $6,380 per member would cut the poverty rate by more than half for people living in larger households. For example, four times a basic grant of $6,380 comes to $25,520, enough to raise almost all families of that size above the FPL.
On the whole, a basic grant of even that modest amount would clearly outperform the present social safety net. According to Robert Greenstein of the Center on Budget and Policy Priorities, about 44 percent of families who would otherwise be poor are currently lifted out of poverty by the benefits they receive.
Work incentives. The preceding calculations for income security are “static” in that they assume a UBI would have no effect on work activity. Many critics worry that a UBI, even at a subpoverty level, would discourage work. They warn of loafers living in yurts in the woods, spending their handouts on drugs. However, a more serious evaluation suggests that only a small minority would choose a lay-about lifestyle.
There is, in fact, a large empirical literature on the ways that increases in income affect work behavior. Some studies are based on how people react to windfalls such as lottery winnings, inheritance, and tribal casino dividends. Others use structural data on wages, salaries, and work behavior.
Typically, such studies do find that, other things being equal, people tend to work less and take more leisure as their incomes increase, but the effect is small.
Studies of the income effect, as economists call it, suggest that other things being equal, a 10 percent increase in income tends to induce a zero to 1 percent decrease in hours worked. For some people, that could mean abandoning employment altogether, but more often, it would just mean taking a vacation or a day off now and then.
Furthermore, in real life, “other things” are not equal. While windfall income has a negative effect on work, an increase in take-home pay from an added hour of work has a stronger and positive substitution effect that increases work effort. The term comes from the incentive to substitute income-producing work for leisure when wages rise.
Estimates are that the substitution effect of a 10 percent increase in take-home pay tends to cause a 1 to 4 percent increase in hours worked.
A UBI that aimed to maximize work incentives could take advantage of the tendency for the stronger substitution effect to outweigh the weaker income effect. For that to happen, the UBI would have to replace the current system of means-tested welfare (a replacement UBI), not be added on top of it (an add-on UBI).
As I have discussed in detail elsewhere, the current welfare system, which aggressively reduces benefits for each dollar earned, seriously erodes work incentives, especially for people just above the poverty line. An add-on UBI would do nothing to offset those disincentives. In fact, if it had even a small income effect, it could actually reduce work effort.
In contrast, a replacement UBI would increase work incentives since the substitution effect of letting low-income workers keep all, or nearly all, of their added earnings would more than offset its income effect. (For a full discussion of the incentive effects of various UBI designs, see here and here.)
Affordability. Any UBI would be costly, in that it would pay benefits not only to the needy, but to everyone. Taxing people with higher incomes and then returning a part of the revenue to them in the form of a cash grant might appear to be a wash, but in fact, taxes are inevitably a “leaky bucket.”
Taxes and transfers not only entail administrative costs; they can also distort economic behavior in ways that undermine growth and efficiency.
The leaky-bucket problem could be lessened if a replacement UBI replaced not only existing poverty programs, but also middle-class income benefits like the tax-deductibility of mortgage payments, charitable contributions, and retirement savings. Since the benefits of those tax preferences are strongly skewed toward upper-middle and wealthy households, a majority of middle-class households would still come out ahead.
The affordability of a replacement UBI could be further enhanced by a “double-dipping” rule that would allow Social Security recipients to take either their current benefits or the UBI, but not both. Restricting the UBI to resident citizens only would further stretch available funding.
However, even with these measures included, I estimate that a replacement UBI could offer a grant of no more than $2,800 to $3,200 per person per year without new taxes. That is only about half of the deep poverty level. Based on static estimates alone, without accounting for enhanced work incentives, such a UBI might well raise a smaller percentage of families out of poverty than current programs.
The safety-net trilemma
Taken as a whole, our discussion of income security, work incentives, and affordability reveals a problem that a UBI shares with other proposed reforms of the safety net – a safety-net trilemma, as we can call it.
For a UBI, the trilemma looks like this: A replacement UBI with a benefit set at half or more of the FPL would arguably outperform the existing welfare system in terms of income security and work incentives, but it would not be affordable without significant new funding.
An add-on UBI with a much lower benefit would be more affordable and could improve income security, but it would do nothing to mitigate the work disincentives of the current system. And a replacement UBI that drew only on existing funding would be affordable and would improve work incentives, but would have a benefit too low to improve income security, and might even reduce it .
The safety-net trilemma is central to the critiques offered by many UBI skeptics. As Greenstein puts it,
If you take the dollars targeted on people in the bottom fifth or two-fifths of the population and convert them to universal payments to people all the way up the income scale, you’re redistributing income upward. That would increase poverty and inequality rather than reduce them.
Other UBI critics provide detailed estimates to support much the same argument. A paper by Hilary W. Hoynes and Jesse Rothstein draws on Census Bureau data to argue that “replacing existing anti-poverty programs with a UBI would be highly regressive, unless substantial additional funds were put in.”
A paper by Melissa S. Kearney and Magne Mogstad reaches the same conclusion. With their views in mind, we turn to forms of cash assistance that aim to address the tradeoffs raised by the safety-net trilemma.
Phaseouts and negative income taxes
One obvious way to address the affordability leg of the safety-net trilemma is to phase out benefits for people with higher incomes. Milton Friedman’s negative income tax (NIT) is the best-known example. Friedman’s original NIT featured a basic grant for those with no earnings, subject to a benefit-reduction rate of 50 cents per dollar of earned income.
Another possibility is a hybrid of a UBI and NIT, which would pay a fixed basic grant to everyone up to an earned-income threshold, after which benefits would be gradually phased out. I will call such a plan a UBI with phaseout, or UBI-PO.
Affordability vs. income security. Adding a phaseout to a UBI decreases total expenditures for any given basic grant. If the basic grant were set at or above the FPL, an NIT or UBI-PO could eliminate all measured poverty. Even if the basic grant were less than the FPL, the UBI-PO does as well in terms of income security as a simple UBI.
Alternatively, we could hold the total cost of benefits constant when comparing a UBI and a UBI-PO. Using a phase-out to reduce the number of beneficiaries and focusing benefits on those most in need would allow the UBI-PO would to award a higher basic grant than a simple UBI. It would therefore have a stronger impact on income security.
Affordability vs. work incentives. The downside of a UBI-PO or NIT, compared with a simple UBI, is that people subject to the phaseout would face reduced work incentives. At best, the phaseout range could be shifted to people whose incomes were already higher and whose attachment to the labor market was stronger, in the hope that the disincentives would be less damaging.
While on the subject of work incentives, it is often noted that a simple UBI financed by a flat-rate income tax would be mathematically identical to an NIT or UBI-PO having the same basic grant and a phaseout with the same rate as the tax. For example, Miranda Fleischer and Daniel Hemel develop that point in detail in a forthcoming paper, “The Architecture of a Basic Income.”
The contention is true as far as it goes, but it implies a rather limited view of public finance.
First, there is no reason either in law or economics that a UBI would have to be fully financed by a single, dedicated revenue source.
Second, even if the UBI were tied to a specific revenue stream, there is no reason it would have to be a tax that was directly a function of income.
For example, the UBI could draw on revenues from a wealth tax, an inheritance tax, a capital gains tax, a consumption tax, a carbon tax, or some combination of the above.
Third, as noted earlier, the cost of the UBI could be offset, in whole or in part, by cutting existing means-tested welfare for the poor or by reducing middle-class tax preferences.
It is a valid point to say that the full effects of any safety-net program should be evaluated in terms of its impact on the government budget as a whole, not on the expenditure side alone. Ideally, government budgeting should evaluate expenditures on their own merit while optimizing the tax system to minimize deadweight losses.
Having done that, broad budget rules should be invoked to establish a sustainable relationship between revenues and expenditures. But those considerations are not specific to income support policies. They should apply to all categories of government spending.
We now turn to wage subsidies, which addresses the safety-net trilemma by attempting to enhance work incentives while improving affordability and maintaining adequate income support. By a wage subsidy, we mean any policy offering benefits that increase with earned income over some range. The existing Earned Income Tax Credit (EITC) is an example.
EITC benefits vary with income according to a schedule that depends on family structure, with very limited events for unrelated individuals and families without children.
For example, as of 2020, a single parent with one child would receive a subsidy of 34 cents for each dollar earned, up to an income equal to 60 percent of the poverty level. In the range from 60 to 112 percent of the poverty level, the EITC benefit reaches a maximum that does not vary with income. From 112 to 240 percent of the FPL, benefits are phased out at a rate of 16 cents per dollar earned.
Work incentives. Although a replacement UBI or UBI-PO could, as discussed above, improve work incentives for people with low incomes, wage subsidies could do even more. Unfortunately, the incentive effects of the existing EITC are substantially diluted by the fact that it is added on top of existing means-tested welfare programs, rather than replacing them.
The benefit reduction rates of other programs thus tend to offset the positive work incentives for people receiving a subsidy and to amplify the disincentive effects on those in the phaseout range.
A 2015 report from the Congressional Budget Office illustrates the problem with several examples, including one based on a hypothetical family of one parent and one child living in Pennsylvania in 2016. In addition to the EITC, the family is eligible for Temporary Assistance for Needy Families, the Supplemental Nutrition Assistance Program, Medicaid, the Children’s Health Insurance Program, and cost-sharing subsidies for health insurance, and is also subject to payroll and income taxes.
Based on underlying data supplied by the CBO, work incentives for this family are strongest in a range from 31 to 54 percent of the poverty level, which lies within the subsidy range of the EITC. In that range, disposable income increases, on average, by $1.33 for each added dollar of earned income.
The weakest work incentives occur in a range from 115 to 127 percent of the poverty level. In that range, the family is subject to the EITC phaseout, as well as to benefit reductions for other programs. As a result, the family keeps only 20 cents in disposable income for each dollar earned.
Income security. From a static point of view, assuming no changes in work behavior, a wage subsidy program alone does nothing to help people who do not work, and does little to help those who work sporadically at low-wage jobs. Its strongest effects on income security are concentrated on the working poor.
However, it seems inappropriate to judge wage subsidies in static terms, since their whole point is to encourage work. The more effective a wage subsidy is in doing so, the more it will improve earnings. In that regard, rate of wage subsidy is not the only aspect of program design that deserves attention.
For example, a wage subsidy would probably have a stronger work incentive if paid out as a monthly cash bonus, rather than as an annual tax refund, as is the case for the EITC. Also, there is some evidence that the substitution effect of a wage subsidy is stronger for second earners in a household where one person already works.
For that reason, it might be worth considering a wage subsidy that applies at least to a limited degree to earnings by a secondary worker, even if the primary earner in the family already earns enough to reach the maximum subsidy.
Finally, in assessing the impact on income security, we need to recognize that the benefits of a wage subsidy do not necessarily all flow directly to workers. Standard labor market economics suggests that at least part of the benefit is likely to accrue to employers.
Suppose, for example, that an employer initially needs to pay $9 an hour to attract the number of workers it wants. If the government adds a wage subsidy of $3 per hour, will employees end up taking home $12 per hour, or will the employer just cut its wages to $6 an hour, leaving take-home pay unchanged? Most likely, something in between. For example, the wage might fall to $8 per hour and take-home pay might rise to $11.
If the goal of the wage subsidy is to increase employment, the tendency for wages to fall is helpful, because it encourages employers to hire more workers. On the other hand, the tendency of wage subsidies to depress market wages does partly offset their effectiveness in enhancing the income security of low-wage workers.
Affordability. As a replacement for the current welfare system, a wage subsidy looks more affordable than either a UBI or a NIT. At the low-income end of the scale, it gives nothing to nonworkers and not much to part-timers or workers whose wages are very low. At the high-income end, the cost of the program is reduced by the phaseout. Most benefits go to those in the middle, including workers who need a helping hand to make the critical step from poverty to stable self-sufficiency.
However, it is harder to imagine a wage subsidy alone as a full replacement for current income-support programs. Doing so would mean no income support at all for people who were unable to work, a politically unlikely outcome.
One possibility would be to maintaining separate programs for those unable to work, but that would possible only if a sharp line could be drawn between those who are unable to work and those who are able to work, but choose not to. In practice, however, a great many people fall into a gray area of “hard to employ” that lies between “able” and “not able.”
The hard-to-employ include people with criminal records, unstable housing, substance abuse issues, family situations that interfere with regular work schedules, borderline mental and physical conditions that fall short of actual disability, and other problems that make it hard to hold a job.
Experience shows that with careful, one-on-one casework, at least some of the hard-to-employ can establish at least a tenuous foothold in the labor market, but the effort required to help them do so also erodes the apparent cost advantage of work-conditioned income support.
The case for integrated cash assistance
Our examination of UBIs, NITs, and wage subsidies shows that there is no way to completely resolve the tensions among the goals of income security, work incentives, and affordability that constitute the safety-net trilemma. Any reform program must recognize the tradeoffs and reach compromises.
Having said that, there is no reason that the ideal set of tradeoffs and compromises lies in a choice of just one of the three canonical policy alternatives. Instead, there is every reason to combine their key elements. The result, which I call integrated cash assistance (ICA), would include the following elements:
- A basic grant that would be paid to all eligible adults, regardless of their employment status, supplemented by a universal child allowance that would not depend on the employment status or earnings of parents.
- A wage subsidy that would provide a bonus for each dollar of earned income to low-wage workers, up to a maximum benefit. The subsidy would be the same for all workers, regardless of family status.
- A phaseout that would reduce the combined basic grant and wage subsidy at a moderate rate, beginning at a high enough level that all families and unattached individuals would be guaranteed a reasonable degree of income security.
A full discussion of the affordability of such a program of integrated cash assistance is beyond the scope of this commentary. Meanwhile, drawing on some earlier calculations based on 2018 data, I can offer some very rough estimates of an ICA that could be fully financed by funds already devoted to existing, means-tested safety net programs:
- An adult grant of about $6,200 per person, roughly half the FPL for an unattached individual. The grant would replace conventional in-kind and cash welfare as well as many middle-class benefits and would be phased out to focus benefits on the low-income half of the income distribution.
- A child grant of $4,500 per child, replacing child nutrition and child-related tax credits, with a phaseout that focused benefits on the poorest half of all children.
- A wage subsidy, replacing the EITC and extended to all adults, with a maximum benefit of about $2,400 for an individual.
The total received by a family would be the basic grant for each adult, plus the child allowance for each child, plus applicable wage subsidies. Benefits would be sufficient to provide an income above the deep poverty level to all unrelated individuals and families even if they had no earned income, and to raise their total disposable income above the FPL if they had at least one adult working at least half-time at the federal minimum wage of $7.25 per hour.
Those are not dramatically generous sums, but even so, it is likely they would do better than Greenstein’s estimate that current programs lift 44 percent of low-income families out of poverty. To the extent enhanced work incentives induced greater labor force participation, the results would be better still.
Naturally, various policymakers would have different ideas regarding which components of the ICA to emphasize, in part because of different weights they assign to the goals of income security, work incentives, and affordability. But the very flexibility of the ICA framework should make it easier to assemble a political coalition to get such a plan enacted.
Finally, it is worth noting that although the ICA would replace many existing safety-net programs, it would not replace them all. The most important separate component of would be health care.
My own preferred design for a separate health care safety net would be some form of universal catastrophic coverage, but there are many alternatives on the table. The important thing is that there be a health care safety net that guarantees affordable access to necessary services for everyone, seamlessly, including those who change jobs or lose jobs in a crisis.
In addition to health care, it would be necessary to supplement an ICA with programs targeted at problems for which cash alone is not a solution. Those would include problems to deal with issues like substance abuse, child abuse, domestic violence, and reintegration of released prisoners into society, to name just a few.
Above all, if there is one thing we have learned from the COVID-19 crisis, it is that our current social safety net is not as resilient as it should be.
The common hazards of life caused by business cycles, natural disasters, or events that strike individuals at random are the known-knowns of social policy — things we are aware of and understand how to deal with. Our safety net is not even capable of dealing with those.
Millions find themselves ineligible for unemployment insurance because they work in the gig economy, or have not been at their present job long enough, or for other reasons. Half the population gets health insurance through their jobs, and when those jobs disappear, health coverage goes, too.
Even moving from state to state to care for a family member can cause loss of unemployment- or health-benefits. We know these gaps exist, but year after year, we do nothing about them.
A truly resilient social safety net should also be capable of dealing with the known-unknowns of life — things we are aware of but do not fully understand. The effects of automation and technological change are an example.
There are many indications that these forces are hollowing out the labor market, with middle-skilled and middle-income jobs disappearing while openings proliferate in low-skill, low-paid sectors, such as home health assistance. Maybe this trend will become a true job apocalypse, as some warn it will, or maybe it won’t.
In either case, a better social safety net would help us cope with whatever technological change throws at us. Wage subsidies to encourage people to fill essential but low-skilled jobs are one idea that could help.
Most important of all, we need a social safety net that is capable of dealing with unknown-unknowns — things that we do not anticipate and do not understand. The COVID-19 pandemic fits that pattern. Not the disease itself — pandemics were a known risk long before this one arrived.
The unknown-unknowns in this case concern the follow-on effects of using social distancing as almost the sole means of controlling spread of the virus. We don’t have a playbook for a sudden, deep recession with this combination of supply shocks and demand shocks.
Having entered the crisis unprepared, policymakers have had to improvise a new social safety net on the fly. With creditable speed, Congress passed the CARES Act, which included something resembling a temporary UBI. Except that it turned out to not to be universal after all.
Officials were soon warning that it would take up to six months to get the cash out to millions of citizens who did not already have their banking data on file with the government, including some of the most needy, such as the unbanked, the homeless, and low-income Social Security recipients who do not file taxes.
The CARES Act also included new supplemental unemployment benefits, but if you lived in one of the 25 states whose computer systems were running on 1980s hardware and software, actually getting your benefits could take a while.
How much easier it would be if the legal and administrative framework for a seamless, work-friendly, and resilient safety net were already in place. A program of integrated cash assistance, as outlined above, would give policymakers a set of buttons they could push and dials they could adjust to expand or modify the system as needed to cope with unknown-unknowns as they unfolded.
Even with no action, those in need would already have at least a small basic grant that they could fall back on in case of illness or job loss. Where that turned out not to be enough, modifications could be introduced without crashing anyone’s software or delays to write new rules.
It is too late now to meet the present crisis with a better social safety net in place, but it is not too late to learn from our mistakes. Who knows, maybe some bits and pieces of what is being cobbled together now can even serve as a framework to build something more permanent to deal with future crises.
Originally published here.