The Case For a Wonky Basic Income Plan

In March, a former union leader joined forces with a prominent libertarian to debate two Obama White House economists. The topic, hosted by Intelligence², was “Is Universal Basic Income The Safety Net of the Future?” with the unlikely duo — Andy Stern, President Emeritus of the Service Employees International Union, and Charles Murray, fellow at the conservative American Enterprise Institute — arguing the case for, and the economists — Jared Bernstein, former chief economist to Vice President Biden, and Jason Furman, former chairman of the Council of Economic Advisors — arguing against. Stern pitched universal basic income — giving every person a check each month without conditions — as insurance for a future where AI and robots make human workers obsolete; Murray lauded the personal liberty from solving poverty without a paternalistic bureaucracy. Bernstein and Furman claimed the math didn’t add up; that the universality requires reducing benefits for the poor, especially families with children; and that other changes needed to pay for the program would be politically infeasible.

From left: Charles Murray, Andy Stern, Robert Rosenkranz (moderator), Jared Bernstein, and Jason Furman.

Economists have expressed skepticism of this idea elsewhere. Bob Greenstein of the Center on Budget and Policy Priorities questions the compatibility of such a large expenditure with today’s political climate. A 2016 survey found that 70% of economists opposed replacing all transfer programs with a basic income of $13,000 per year for adults age 21 and over. Others like blogger Tyler Cowen believe that work, not money, is the central problem for the country.

And yet, there’s a lot for economists (and us all) to like about basic income. It

  • Fills in the gaps of the current safety net, which often fails to reach the poorest and unemployed.

Their concerns are often not about basic income per se, but about the plans proposed. Political challenges aside, economists’ most compelling critique may be that existing plans harm those in most need, especially families with children. Bernstein and Furman drilled this point in the Intelligence² debate, leading them to victory: while 20% of the audience initially opposed the motion that “the universal basic income is the safety net of the future”, 61% opposed it afterward.

The good news is: it’s entirely possible to design a basic income plan that minimizes the net losses (and therefore political opposition) to beneficiaries and taxpayers, while still respecting the core tenets of basic income — universality, individuality, and focus on cash. Doing so requires a person-centric perspective, whereby we consider the current benefits received and taxes paid by all Americans today, and design the plan with the explicit goal of giving as many people as possible at least as much as they currently get.

This would be hard work, and I’ve not (yet) formed a specific plan. This piece lays out the justification for a new approach to basic income policy design, describes a framework for this design, and proposes negative income tax as a conceptual starting point. Such a policy can win expert support for basic income by emphasizing welfare of individuals.

The current plans

Stern and Murray were appropriate defenders against Bernstein and Furman, as they have proposed some of the most specific and widely-cited plans for basic income in the US. Prominent basic income advocate Scott Santens also released a plan in June. So it’s worth summarizing the Stern, Murray, and Santens plans on a few dimensions:

  • Annual amount
    Stern: $12,000
    Murray: $10,000, plus $3,000 to purchase catastrophic health insurance
    Santens: $13,266 for adults, $4,598 for children

Current proposals reduce benefits for some

By limiting to adults, the Stern and Murray plans both leave most poor single-parent households significantly worse off than they are today, as Daniel Hemel finds in his well-researched criticism (this also describes the “plan” in the survey question opposed by 70% of economists). These households currently receive thousands of dollars per year from EITC, CTC, SNAP (f.k.a. food stamps), and other programs like housing assistance (if they’re in the lucky 24% of eligible households to get it).

Santens proposes $4,598 per child, in accord with federal poverty guidelines (adding 10% to compensate for a VAT), which vary depending on household size. But does even this amount per child compensate for lost benefits and higher taxes? Not for everyone. Consider two households earning $10,000 per year: one with a single childless person, and the other with a single person and a child. The value of marginal benefits received by the latter household is attributable to the child.

The three largest programs for this sample household total $5,958 in value, or $1,400 more than the basic income:

  1. $3,002 from EITC ($3,371 vs. $371 for the childless household)

“In most of the [basic income] proposals, the losers tend to be households with more children.” — Jason Furman

Another example: all three plans are limited to citizens, despite some existing programs being partially available to legal permanent residents.

Plans’ various tax increases create more losers

In addition to some households getting less in benefits under the proposed plans, the suite of tax increases required to finance them would hit virtually everyone, including the poor. To name a few items, people could expect to pay more for:

  • Gas and electricity (carbon tax)

This doesn’t mean these taxes aren’t worthwhile on their own: each enjoys broad support among economists (I support all).

But these ideas bear no innate relationship to basic income, and their political baggage would weigh down the cause. For example, economists have been trying to kill the mortgage interest deduction for decades, only to be consistently rebuffed by the real estate lobby. Pursuing basic income need not require battling the real estate, agriculture, and fossil fuel industries.

It’s also hard to pin down how much each person pays for these taxes, relative to, say, income tax hikes. That increases the variance of who wins and who loses under these plans, meaning they could hurt even more low-income households.

For reasons I elaborate on shortly, I favor income taxes instead to avoid these issues.

A person-centric direction

Fundamentally, the three plans start with a number: the amount believed needed to fund a basic income for the included population. These are big numbers — roughly $3T per year for each plan — so then they fill in the gaps with whatever it takes (except higher income taxes). Examination of example households has followed the design.

Let’s turn that on its head: instead of starting with the number, we can start with the people, and focus on maximizing the upside of each while minimizing the downside. The units of analysis would include people and households, not just dollars and cents.

This paradigm is harder to work in. It requires deep understanding of household income distributions, safety net program design, and tax codes. It then requires model-building to alter each while minimizing negative impacts. This often involves teams of policy analysts under the direction of established think tanks or policymakers. So it’s understandable that Murray, Stern, and Santens haven’t gone there yet. While I don’t yet have a plan that does, I have some ideas on how to build one.

How to build a person-centric UBI plan

We can distill policy design to three pieces:

  1. Non-negotiables. What’s the heart of the policy?

For basic income, non-negotiables (1) are fairly straightforward: a basic income plan must be universal, unconditional, and 100% cash. Universality may be stratified, such that children, people with disabilities, seniors (who receive Social Security), and/or noncitizens receive different amounts. It may not be stratified by the composition of the household; for example, two adults must receive the same amount whether they are in the same household or not (it is an individual-level program, in contrast with many existing programs).

Levers (2) could include several things, starting with those implicit in the three cited plans:

  • Amount per adult (the primary lever)

Assessment (3) should concern the distribution of impacts to after-tax after-transfer income to individuals or households. This is difficult in the Murray, Stern, and Santens plans, since some of the funding sources aren’t easily translatable to this metric, but the one-parent one-child household example suggests that many households with children will see considerable drops in total income.

The most straightforward metric for quantifying this would be something like average % loss per person, i.e. ignore the winners and focus on losers. So if a two-person household was getting $30k after taxes and transfers (including in-kind benefits like SNAP), and with UBI they get $29k, their contribution to the metric would be -3.3% ($1k / $30k), weighted for two people.

This could be extended in a couple ways. For example, winners could be included, though with less weight since politics is (appropriately) loss-averse. Studies in behavioral economics suggest that “losses are twice as powerful, psychologically, as gains,” so losses could receive double the weight of gains. The metric could also consider implicit wages lost due to program compliance time — perhaps at minimum wage — given existing programs require more paperwork and in-person time with bureaucrats than cashing a basic income check does. One could imagine other factors as well, such as utility lost due to restrictive in-kind benefits (Pareto efficiency and welfare economics could be other angles).

Within this framework, the goal becomes: ensure some level of universal cash transfer by adjusting levels and financing, such that negative impact on households is minimized.

Negative income tax as a starting point

The simplest way to achieve this goal is to mimic the current suite of programs, with three important changes:

  1. Replace in-kind benefits with cash.

Since existing programs are means-tested instead of universal, this would not produce a true basic income. What it would produce is what’s called a negative income tax, whereby people below a certain income level receive a payment which phases out with income. It turns out that most historical studies on “basic income” have actually been on negative income tax, which nearly became law in the 1970s.

Importantly, one could design a negative income tax with identical outcomes to a universal basic income, and vice versa, once you set the baseline. The simplest way to turn negative income tax into basic income is to calculate the extra amount individuals get (anyone with income gets more under basic income than negative income tax), and raise their taxes by exactly that amount. After taxes and transfers, nobody gets any more or less. I’ve written more about this, including reasons to still prefer basic income to negative income tax.

To make this a bit more concrete, consider a household with one adult and one child. As FiveThirtyEight’s Andrew Flowers showed in the chart below, this household faces “welfare cliffs” due to immediate or rapid phase-outs of existing programs like TANF, Medicaid, and CHIP, as well as the way they interact with each other and with income taxes.

Earnings vs. after-tax after-transfer income. Source: FiveThirtyEight (line added by Max Ghenis).

The household gets ~$19k when earning $0, and ~$38k when earning $50k, or an average implicit marginal tax rate of 100%-($38k- $19k) / ($50k-0) = ~60%. If that seems harsh, it is: while income taxes are progressive, the highest marginal tax rates (including assistance) are actually borne by low-income households, due to means testing.

I’ve added an approximate line of best fit to this chart to show what this would look like as a flat income tax. We can implement this line in one of two ways:

  1. Negative income tax with a $19k minimum (for the two-person household), 60% phase-out (up to $30k, at which the household neither gets a refund nor pays taxes), and 60% tax rate after $30k (up to $50k earnings)

These produce the same result. The household gets/pays the same amount given their earnings, and the government gets/pays the same in net taxes. The primary difference is that negative income tax comes without the sticker shock: Wiederspan, Rhodes, Shaefer (2015) estimate that a negative income tax set at US poverty line and with a 50 percent phase-out rate could be 95% paid for with the combined budgets of EITC, SSI, SNAP, TANF, school meal programs, and housing subsidies (here’s a Vox summary).

Here lies the core advantage of funding basic income with income tax increases: you can only make a basic income plan equivalent to negative income tax when funding via income taxes. The increase that may seem significant would not actually be felt by households, and with income taxes representing 47% of federal tax revenues, it’s a known quantity that doesn’t create new enemies.

A person-centric analysis will reveal the appropriate minimum and shape of the line to minimize losses for households. This will depend partially on what programs basic income replaces (I tend to agree with Stern and Santens in leaving out healthcare programs like Medicaid and CHIP). Inputs to the analysis would include distributions of household earnings by size, and the inclusion criteria and amounts of each existing program (essentially the above chart for each household type). Turn the resultant negative income tax into a basic income, and we’ve got a fair, focused, realistic plan.

Idealism attracts people; realism convinces them. Media coverage of pilots and automation have generated massive interest in the idea of basic income, but without a better-thought-through plan, that interest is doomed to evaporate among those concerned with the welfare of real people.

It’s time to get in the weeds and make a plan that not only lives up to the ideals of basic income, but also maximizes the number of people benefiting from it.

Economist. Founder and president of the UBI Center. Studied at MIT and UC Berkeley. YIMBY. Former Google data scientist.