Eliminating poverty seems like an impossibly utopian goal, but it’s actually pretty easy: we can just give people enough money that they’re above the poverty line. That idea, known as a basic income, has been around forever, but it’s made a comeback in recent years.
And it’s a sign of how far it’s come that opponents of the idea are beginning to feel the need to make arguments against it. Pascal-Emmanuel Gobry, in The Week, is the latest to present a case against, and grounds it almost entirely in the findings of a series of experiments on a variant of the basic income known as a “negative income tax” conducted in the 1970s, which he says show the idea is doomed to failure.
Not so fast — the experiments raise valid worries, but they hardly herald doom, and still suggest that a negative income tax could eliminate poverty at a manageable cost.
The 1970s experiments
A negative income tax isn’t precisely the same thing as a basic income, but it’s related: after giving everyone a cash grant, an NIT rapidly taxes it away, such that the vast majority of taxpayers get no money back at all. For example, Richard Nixon, during his first year as president, proposed a negative income tax that would pay around $10,000 in 2014 dollars to a family of four, and then tax it away at a 50 percent rate until families earning above $20,000 or so stopped getting anything at all.
The four experiments Gobry cites — conducted in New Jersey, Pennsylvania, Iowa, North Carolina, Seattle, Denver, and Gary, Indiana, with samples totaling about 7,500 people — are, along with a similar experiment done around the same time in Manitoba, the most comprehensive tests to date of negative income taxes. They tried various cash grant sizes (from 50 percent of the poverty line to 148 percent) and phaseout rates (from 30 percent to 75 percent), enabling a more detailed look at how the plan’s components interacted with each other.
The studies found that the policy was beneficial to those getting the money, but tended to modestly reduce the number of hours they worked, and the amount they earned. The latter is a potential cost worth weighing against the policy’s benefits. But to Gobry, it’s definitive proof the plan is defective. “Millions of people who could work won’t, just listing away in socially destructive idleness (with the consequences of this lost productivity reverberating throughout the society in lower growth and, probably, lower employment, in a UBI-enabled vicious cycle),” Gobry concludes.
The problems with concluding too much
The Old Miller School in Gary, Indiana, one of the sites of the negative income tax experiment. (Chris Light/Wikimedia)
Gobry is right that the negative income tax experiments are the best test we have of this policy to date. But “best” does not equal perfect. My concern is that Gobry reads the experiments to be saying more than they are in fact saying, given both flaws and limitations in their methodologies and other conclusions they came to that Gobry failed to mention. Here are a few concerns worth raising.
1. “Worked less” sometimes means “the results were underreported.”
This is the big one. Brookings’ Gary Burtless, writing up the results, noted that the Gary and Seattle/Denver experiments relied on self-reported earnings information, rather than using official government records. When the findings were cross-referenced with actual earnings data, the labor force effects in Gary disappeared entirely, and the Seattle/Denver ones were diminished considerably.
As Princeton’s Orley Ashenfelter noted in a response to Burtless, this throws the entire conclusion that negative income taxes reduce labor supply into jeopardy. “Who is to say whether there would be any labor supply response, further income underreporting, or neither, if an experiment with conventional administrative procedures were implemented?” he asks. “Only an experiment fully informed at the design stage about the possibility for income underreporting, and that tested for its effect, would shed any light on this critical issue. Sadly, the design of none of these experiments was so informed.”
2. “Worked less” does not necessarily mean “dropped out of the labor force forever”
Let’s assume, for the sake of argument, that underreporting doesn’t invalidate all these results. That still doesn’t mean the experiments are the slam-dunk case against basic income Gobry takes them to be.
For one thing, it’s worth differentiating different ways that labor supply can fall. The most obvious way is that people will drop out of the labor force entirely. But as Georgetown philosopher Karl Widerquist, a vocal advocate for a basic income, noted in his write-up of the experiments, researchers didn’t find any evidence that happened. “Would a large number of people respond to an NIT by withdrawing entirely from the labor force?” he asked. “The experiments found no evidence of such behavior. Some of the experimenters said that they were unable to find even a single instance of labor-market withdrawal.”
So what happened, then? Burtless reported that the apparent decline in labor supply didn’t come primarily from a reduction in hours worked either. It’s not that people who had previously been working 60 hour weeks waiting tables cut back to 40 hours.
The remaining explanation, once we’ve ruled out reduced hours and permanent labor force drop-outs as major factors, is longer spells of unemployment. That has obvious costs; unemployment is generally bad for one’s well-being. But the key here is that the negative income tax resulted in people choosing to remain unemployed for a longer stint; presumably, this is more pleasant than involuntarily elongated unemployment.
Further, the most obvious interpretation is that people are waiting longer to find a good job match, or are quitting bad jobs in favor of searching for better ones. Those responses have efficiency advantages and, in the long-run, connecting people with more pleasurable and rewarding work should increase well-being, something worth weighing against the well-being cost of increased unemployment.
3. “Worked less” sometimes means “got more education”
Another factor is people withdrawing from the labor force to pursue more education. Stanford’s Eric Hanushek, evaluating the non-labor force effects of the experiments, found that “for youth the reduction in labor supply brought about by the negative income tax is almost perfectly offset by increased school attendance.”
That’s not the only positive education finding. One study looking at the New Jersey experiment found that a negative income tax of mid-range generosity increased odds of completing high school by 25 to 30 percent; a similar analysis of the Seattle-Denver experiments put the number at 11 percent. While the evidence on academic performance was more limited, there was some evidence that children in NIT households did better at standardized tests in lower grades.
4. “Worked less” is sometimes a good thing
Apart from the special case of education, it’s worth asking, in general, whether maximizing labor force participation is actually a good thing, or whether there’s more to human flourishing than just that. This is not to say that work is unimportant. Gobry notes research suggesting that unemployment comes at a significant well-being cost, and while some of that is probably due to financial stress more than a lack of psychic fulfillment from work, the latter factor is part of the picture.
But especially in cases where people are choosing not to work, it’s worth asking whether they’re being irrational, and setting themselves up for unhappiness, or whether they’ve actually identified something besides work that create even more value in their lives. The whole point of Social Security, for instance, is that at some point in one’s life having a leisurely retirement is better for well-being than continuing to work. Social Security and private retirement savings almost certainly have a much more negative effect on labor force participation than a negative income tax would, and yet we all, correctly, have decided that’s besides the point.
So it’s worth considering whether some of the decrease in labor force participation hypothetically caused by a negative income tax would be desirable. What if a household uses the money to quit a part-time job in favor of caring for a chronically ill child? What if they use it to retire a few years early? What if they use it to finance a long vacation between jobs? I don’t actually think any of those phenomena are big problems policymakers should be eager to avoid.
5. You can only know so much from short experiments
The final caveat is that extrapolating from short-term, small-scale experiments like these to determine the effect of a national or state-wide policy would have is fraught with peril. The experiments didn’t apply to everyone in the municipalities in question; they were targeted at small, low-income subsets of the population. Would the same results hold if everyone were getting the check? You could imagine a permanent plan either having a more positive impact, by triggering macro effects like employers bidding up wages to convince people to stay in the labor market, or a more negative one, by assuring people that staying out of work longer is going to be viable long-term. We don’t know which of those would actually occur.
Scaling an idea like this up from pilot to actual program introduces a wide array of confounding factors. When you find big effects, then it’s plausible those confounding factors won’t be enough to make them go away. But with results as small as the experiments’ findings on labor supply (or the findings on education, to be fair), it’s likely they’ll be overwhelmed by these new complications, and the impact of a national program will look significantly different from what was found at the hyper-local level. As Gobry says, randomized controlled trials (RCTs) like these experiments are among the best tools policymakers have. But what they give us in scientific validity, they lose in scope.
What we should be debating
Lyndon Johnson on his “poverty tour” of Appalachia, pegged to the start of the War on Poverty. (LBJ Presidential Library)
When researchers come to small and somewhat sensitive-to-error conclusions about a policy’s effects, the findings become a bit of a Rorschach test, with interpretations reflecting the policy preferences of those doing the interpreting (myself very much included) more than they do the underlying facts.
“The prevalence of small effects opens the way to alternative interpretations of the research findings,” Nobel economist Robert Solow, commenting on the results, wrote. “The interpretation adopted will depend a lot on the interpreter’s ideological and doctrinal preconceptions and only a little on the detailed experimental results themselves.
But one finding we can rely on, with some degree of confidence, is the conclusion of Burtless and others who have evaluated the negative income tax experiments that a national NIT big enough to eliminate poverty isn’t budgetarily unviable.
Widerquist notes that studies estimating the cost of compensating for lost earnings could increase the cost of the program 5 percent (at the low end) or nearly triple it (at the high end). Burtless, who produced the high-end estimate, concluded that a generous plan set to 100 percent of the poverty line would all the same only cost about 1.5 percent of GNP (which is basically the same as GDP where the US is concerned) a year on top of existing welfare programs. That’s a rough estimate, especially give how much welfare programs have changed since Burtless was writing — Demos’s Matt Bruenig has more current numbers here — but going from the federal government being 21.1 percent of GDP to 22.6 percent or thereabouts is hardly a sea change. And yet that’s, roughly, all it would take to eliminate poverty in America.
So here’s my takeaway: a negative income tax or basic income of sufficient size would, by definition, eliminate poverty. We still don’t know if there’d be much of a cost in terms of people working and earning less. If there is, the effect is almost certainly small enough that a negative income tax can offset the lost earnings and remain affordable. The worst case scenario is that we eliminate poverty but see a modest decline in employment. The best case scenario is we eliminate poverty at even lower cost and don’t see much of an effect on employment. That’s a gamble I’m willing to take.