The Friedman-lite welfare state and how to make the most of it
We never got a negative income tax, but its logic reshaped social policy. To get further, though, we need more than economic reasoning.
A hallmark of the Niskanen Center’s approach to social policy is the idea that free markets and welfare states are complementary forces that provide a foundation of stability and prosperity for the countries that adopt them. But some welfare states are designed better than others – and some are not designed well at all.
Milton Friedman clearly understood the benefits of free markets. While he may not have seen the welfare state as complementary, he would have agreed that a badly designed welfare state can undermine stability and prosperity. As Jennifer Burns chronicles, Friedman’s concern with some of these badly designed policies led him to advocate a negative income tax (NIT), most famously in his 1962 book Capitalism and Freedom. A complex array of anti-poverty programs peppered the federal landscape at the time, and an even more complex array of service-based proposals would emerge as part of Lyndon Johnson’s “war on poverty.” The NIT offered a simple and efficient substitute that could reduce poverty while also minimizing market distortions.
By replacing existing programs and regulations with a single, unrestricted cash benefit for low-income households that would be gradually phased out as household earnings increased, the NIT offered a bold solution that theoretically met the demands of policymakers across the political spectrum. It was anti-poverty, pro-work, and pro-market. But despite the intellectual splash it made at the time, there was very little political success for Friedman’s idea in the years following Capitalism and Freedom’s publication.
As Neil Gross explains in his essay for this forum, academic policy history tends to focus on the rise and fall of President Nixon’s Family Assistance Plan (FAP). Channeling Brian Steensland’s excellent book, Gross explains how the cultural categories of “deserving” and “undeserving” poor proved fatal to the FAP’s negative income tax-style design, which collapsed the distinction between those who were working or were unemployed for reasons beyond their control (such as disability or age) and those who were unemployed but otherwise able-bodied. The former were viewed as deserving of assistance. The latter were, almost by default, assumed to be undeserving, so it was thought that government assistance, if it was provided at all, should come with onerous work requirements to prevent shirking.
It is a convincing history for explaining the failure of NIT-style programs to take hold in the United States, but it is incomplete if the implication is that this makes us exceptional. Despite the global reach of Friedman’s idea, the NIT as he envisioned it has failed to take hold in any rich democracy. The simple reason is that most countries – particularly the Anglo countries inheriting England’s Poor Law Legacy – all have important cultural distinctions between the deserving and undeserving poor that are reflected in their social policies. The more interesting story is how modified versions of Friedman’s NIT idea have fared for the groups universally deemed deserving, including children.
Given the failure of NITs to take hold anywhere (aside from countless small pilot programs), we are better off looking at the trajectory of what might be called Friedman’s trifecta. Instead of consolidating an array of social programs into one NIT, policymakers have chosen to shrink traditional social assistance; consolidate child benefits into more generous refundable child tax credits (like the temporary expanded credit American families got during the pandemic); and smooth the transition from welfare to work with in-work tax credits for low-income workers (like the Earned Income Tax Credit).
We see the rise of the Friedman trifecta in the 1990s and 2000s across most liberal welfare regimes, including Canada, Australia, New Zealand, and the United Kingdom, which all adopted welfare reform, in-work tax credits, and consolidated child tax credits. The United States remains an outlier even within this universe, however, because of the limited nature of our Child Tax Credit: It is the only such program with a minimum earnings requirement. Apart from a very short moment during the pandemic (discussed below), American families have always had to meet an earnings threshold to receive the full CTC.
In an academic article aptly titled “Friedman’s Revenge,” John Myles and Paul Pierson explain how the initial failure of NIT-style reforms in the 1960s was followed by some success after 1974 in the United States and Canada. The story will be familiar to those who follow Pierson’s work on changing welfare states. The oil crisis brought an end to the welfare state’s golden age and ushered in a new era of austerity. In this context, universal social policies became increasingly unsustainable from a fiscal perspective. Adding some element of income-testing to previously universal social policies, such as flat-rate pensions and family allowances, became an attractive way to increase benefits for lower-income households without increasing deficits. Countries also opted to introduce targeted — and much less expensive — in-work tax credits in lieu of other program expansions as a way to “make work pay” for low-wage households. If Friedman was the intellectual father of the NIT, austerity is its mother. “NIT programs,” according to Myles and Pierson, “are a progeny of austerity.”
“Friedman’s Revenge” revisited
The United States and Canada were directionally similar in the shift toward NIT-style policies, but there were important differences between the two countries at the specific program level. This is most evident in family policy. Myles and Pierson saw Canada hewing closely to Friedman’s vision with the decline of traditional means-tested social assistance and universal family allowances and their replacement with generous, refundable child tax credits that phase out for higher income families. In the United States, the decline of Aid to Families with Dependent Children (AFDC) was partially offset by the rise of the Earned Income Tax Credit (EITC), but there was no significant compensation in family policy.
Three factors explain this divergence, according to Myles and Pierson. With its parliamentary system, Canada had fewer veto points to block reform. With no analogue to the U.S. South, Canada had less intense opposition to reform. Finally, Canada’s already existing, universal family allowances provided a bridge to NIT-style reform because they could be easily converted to an income-tested program. It is a classic example of the effective use of comparative-historical analysis to understand policy trajectories in two otherwise similar countries.
Nonetheless, it is worth revisiting Myles and Pierson’s central theses in light of more recent history. Two developments are particularly important – one historical and one intellectual.
First, just as Myles and Pierson’s original piece went to press, the United States was on the verge of introducing a very expensive child tax credit (CTC) as part of the Taxpayer Relief Act of 1997. The new CTC was worth $500 per child and did not begin phasing out until well above the median income for American families. Within a few years, Congress would double it to $1,000 per child. (Unlike its Canadian counterpart, the American CTC was nonrefundable, phasing in and excluding the lowest-income families altogether.) Shortly thereafter, Canada introduced an in-work tax credit similar to the EITC. In both countries, the wholesale creation of new tax credits was inconsistent with Myles and Pierson’s emphasis on the cost-saving rationale for the shift to child and in-work tax credits.
Second, the rise of fiscal sociology as a subfield pushed us to rethink how we conceptualize tax expenditures (credits, deductions, and exemptions) as social policy. Myles and Pierson consider refundable tax credits to be part of both countries’ policy legacies. But they ignore the United States’ historical reliance on a different mechanism — tax exemptions for dependents — to provide benefits for families with children. Whereas tax credits shrink the taxes owed at the bottom line, exemptions shrink the amount of income subject to tax.
In the United States, families with children have been able to knock thousands off their taxable income for as long as there has been an income tax. This was a universal policy of family support that could have easily served as a bridge to NIT-style reforms. Policymakers would have merely needed to convert the exemption to a refundable child tax credit that would be subject to some income-testing to target lower- and middle-class families.
But it didn’t happen that way. Instead, the U.S. child tax credit has excluded the poorest children from its inception.
The CTC and the logic of tax relief
Conceptualizing dependent exemptions as a type of family allowance analogous to Canada’s suggests we should have seen U.S. lawmakers at least try to build on that policy legacy.
That is exactly what we saw beginning in the 1970s. Because the federal income tax was not indexed for inflation in the post-1974 stagflation era, inflation was eroding the real value of tax exemptions and bracket creep was pushing households into higher tax brackets. The result was rising taxes on families. This is part of the reason why Reagan’s 1981 tax cut, which included provisions to index the tax system and reverse years of backdoor tax hikes, was so popular. The following year, the American Enterprise Institute asked Eugene Steuerle, then a Treasury official in the Reagan administration, to write an essay on the taxation of families. He found that inflation had cut the value of the dependent exemption in half since 1948. This meant that rising taxes had hit families with children especially hard. This research was widely discussed in the press, with the New York Times running a story under the headline “Substantial Rise Found in Federal Tax Burden on the Poor.” Rising family tax burdens called for family tax relief. Heeding Steuerle’s advice, Reagan subsequently made doubling the dependent exemption (and increasing the EITC) a major component of the 1986 tax reforms.
In other words, the dependent exemption in the Reagan era became defined solidly as a program of taxpayer relief — a development that would stymie efforts to extend it to all families in the future.
Policy advocates certainly tried to make the dependent exemption work as a bridge to a fuller Friedman trifecta. Steuerle himself shifted from his early focus on expanding the dependent exemption and began advocating its conversion into a universal child tax credit. Building explicitly on his earlier work, he co-authored a 1991 white paper with Jason Juffras that took the next logical step in completing the Friedman trifecta. The white paper helps us see the path not taken by American policymakers in the 1990s and pinpoint the reasons why. Steuerle and Juffras noted that most families with children already received cash benefits as either monthly AFDC checks or yearly tax refunds when they claimed dependent exemptions or the EITC. They were good economists like Friedman and recognized the inefficiencies built into the system at that time.
Echoing Friedman’s case for an NIT to replace existing programs in the 1960s, they argued the status quo was unfair (“Children with equal needs do not receive equal treatment in the tax, welfare, and health systems, violating a basic principle of fairness”) and often acted as a barrier to upward mobility (“The welfare and tax systems together create strong disincentives for low-income parents to work, marry, or move to find a job”).
As we saw in Canada, their proposed reform was to use the dependent exemption as a bridge to a refundable child tax credit with no earnings requirement:
Programs that duplicate the child credit would be curtailed. The dependent exemption for children would be eliminated. The Aid to Families with Dependent Children (AFDC) program would be cut by over one-third, reducing the strong disincentives of this program on work and marriage. Social Security benefits for non-disabled children of Social Security retirees would also be eliminated… The child credit would be indexed to grow in line with the new sources of revenues, most of which will rise faster than inflation. Thus, by 1996 the credit would equal approximately $1,600 in current dollars and about $1,100 in 1990 dollars. As the real value of the credit increased, it would continue to displace AFDC appropriations.
The proposal was picked up by the National Commission on Children and included in its landmark final report, Beyond Rhetoric: A New American Agenda for Children and Families. In the minds of economists, the dependent exemption was the perfect bridge to a fully refundable child tax credit for the usual reasons — it reduced the overall cost of reforms, streamlined complex benefits, tackled child poverty, and reduced barriers from welfare to work. From an economic perspective, it should have been an easy policy choice.
Except it was a complete political failure. A bill sponsored by Senator John “Jay” Rockefeller, who chaired the National Commission on Children, failed to attract co-sponsors. Liberal Democrats introduced several less ambitious refundable CTC bills — all of which quickly sputtered out for lack of support among the broader Democratic caucus.
The limits of thinking like an economist
The problem facing efforts to convert the dependent exemption into a refundable child tax credit is the same one facing efforts to convert our maze of social programs into a broader NIT as Friedman envisioned. Although such reforms usually make a lot of sense from a narrow economic perspective, they fail to account for the institutional logics that influence how we perceive the appropriateness of certain policy reforms relative to others.
In this case, what economists perceived as a perfectly fine economic bridge was perceived by policymakers as an inappropriate cultural bridge. In one hearing, for example, Rep. Dennis Hastert cautioned his colleagues, “It’s a fine line between family tax policy and welfare reform. We need to be careful not to get the two mixed up.” Writing in response, Family Research Council president Gary Bauer replied, “I fear some want to increase welfare under the rubric of pro-child tax relief” and went on to argue the case for “separating tax relief from cash transfers.” Dependent exemptions and child tax credits that phase in are a feature — not a bug — when the goal is to ensure that tax burdens do not push families into poverty, as had happened in the years following 1974. In this schema, families not earning enough to pay income and payroll taxes are undeserving of tax relief even if they are deemed deserving of some temporary assistance through welfare programs.
Congressional Republicans understood this distinction much better than Democrats. While Democratic bills languished, conservatives pushed for the introduction of a partially refundable child tax credit that phased-in with earnings. Writing around the same time as Steuerle and Juffras, Robert Rector and Stuart Butler of the conservative Heritage Foundation proposed, “A practical way to provide reasonable tax relief to these low-income families would be to replace the current dependent exemption with a partially refundable ‘child credit.’ Parents could use this credit to reduce both their income tax and the employer and employee Social Security tax liability.” Rector and Butler were careful to distinguish their proposal from Steuerle-style proposals that provided “welfare benefits to non-working families enrolled in the Aid to Families with Dependent Children (AFDC) program.”
This approach was exceedingly successful precisely because the public and elites could see continuity between the dependent exemption and the proposed tax credit — both of which pursued a narrower goal of family tax relief. Republicans made this version of the child tax credit the “crown jewel” of their famous Contract with America in 1994. Congress ultimately passed a nonrefundable child tax credit in 1997. Three decades later, Republicans finally succeeded in combining the dependent exemption with the child tax credit to create a partially refundable $2,000 child tax credit as part of the Tax Cuts and Jobs Act of 2017. The dependent exemption was the bridge for which Myles and Pierson had been looking — only it did not necessarily lead to universal, NIT-style reforms, as they might have imagined.
Using our sociological imagination
One of the classic sources of institutional change is the much-touted exogenous shock. In this model, policy continues along well-worn grooves until a large shock from outside the system temporarily creates an opportunity for a big divergence. For many, it looked like the COVID-19 pandemic was the exogenous shock that might lead to major changes in American social policy.
Working with Congress, the Trump administration passed a bold series of economic supports to businesses and families to help them deal with the pandemic’s economic fallout. This included the country’s first fully refundable tax credit for children. President Trump prominently displayed his name on the Economic Impact Payment (aka stimulus) checks that went out to families and their children. The Biden administration built on Trump’s earlier round of checks by temporarily increasing the value of the child tax credit and making it fully refundable for the first time. Advocates capitalized on the expansion, soon amassing studies showing that the poverty impacts were large, employment effects were small, and we could finally be free of the credit’s legacy of phasing in with earnings.
Like their economist predecessors in the 1990s, they turned out to be very wrong. Eliminating the CTC’s phase-in was a relatively minor tweak from a technical perspective but a major departure from the credit’s history from an institutional logics perspective. The logic of making the CTC temporarily fully refundable was more akin to the thinking behind the stimulus checks — a temporary response to an acute crisis — than the acceptance of the totally new reasoning of income supplementation that had undergirded support for child benefits in other countries for decades. When the crisis faded, so did support for Biden’s expanded child tax credit, opening it up to the same non-economic critiques that held back expansion prior to the pandemic.
Where does that leave advocates of the Friedman trifecta who would still like to see a fully refundable child tax credit to complement changes to the EITC and TANF? Once again, we must look beyond economics to political science and sociology to find the answers. Historical institutionalists have described several alternatives to the exogenous shock model of policy change. In these models, big structural change emerges slowly through what might be called relentless incrementalism. The most promising path for policy change is via the displacement of the dominant policy logic, not its sudden implosion. Rather than aim for a radical reconstruction of the thinking behind the child tax credit, policy entrepreneurs should cultivate subordinate or new lines of thinking that have historically existed alongside dominant logics.
On the federal level, Senator Romney’s Family Security Act builds on recent conservative innovations, including consolidating overlapping tax benefits, reducing work and marriage penalties, and supporting traditional families — all of which were important considerations in previous CTC reforms, where they played secondary roles. By elevating these ideas over traditional concerns about tax relief, the proposal could begin to displace the dominant logic and open up new possibilities for reform.
Meanwhile, states are experimenting with new innovations in child tax credits in the wake of the 2017 and 2021 reforms at the federal level. The logic of these state tax credits is less well-defined and more open to contestation. Eleven states have introduced fully refundable child tax credits in recent years. They come in all shapes and sizes, but dominant models are likely to emerge and diffuse across states. In a federal system, state innovations often feed back into federal innovations. Wolfgang Streek and Kathleen Thelen write: “As growing numbers of actors defect to a new system, previously deviant, aberrant, anachronistic, or ‘foreign’ practices gain salience at the expense of traditional institutional forms and behaviors.” Just as most New Deal reforms had their origins in existing state policy — think mother’s pensions or Wisconsin’s unemployment insurance — state CTC innovations will eventually impact federal reforms.
All this is to say Friedman’s negative income tax remains a worthy ideal — but the economic thinking that gives it strength is also its biggest weakness. By considering its political underpinnings and the sociological obstacles it faces, we can still muddle our way toward a second-best Friedman trifecta and give him the credit he deserves for improving our world.
Joshua T. McCabe is Director of Social Policy at the Niskanen Center and the author of The Fiscalization of Social Policy How Taxpayers Trumped Children in the Fight Against Child Poverty (Oxford, 2018). He tweets @JoshuaTMcCabe.